Search is not available for this dataset
pmcid
large_stringlengths 8
11
| pmid
large_stringlengths 0
9
| authors
large_stringlengths 0
33.4k
| title
large_stringlengths 0
2.49k
| publication_date
large_stringlengths 10
10
| keywords
large_stringlengths 0
17k
| abstract
large_stringlengths 0
122k
| text
large_stringlengths 9
8.39M
|
|---|---|---|---|---|---|---|---|
PMC9647015 | Cankun Zhou,Chaomei Li,Yuhua Zheng,Xiaobin Huang | Regulation, genomics, and clinical characteristics of cuproptosis regulators in pan-cancer | 27-10-2022 | cuproptosis,pan-cancer,genomics,tumor microenvironment,immune | Background Cuproptosis, a copper-dependent controlled cell death, is a novel form of cell death that differs from known cell death mechanisms; however, its overall regulation in cancer remains elusive. Methods Multiple open-source bioinformatic platforms were used to comprehensively elucidate the expression levels, prognostic efficiency, potential biological functions, genomic and epigenetic characteristics, immune microenvironment, and drug sensitivity of cuproptosis regulators (ATP7A, ATP7B, DLAT, DLD, FDX1, GLS, LIAS, LIPT1, MTF1, NLRP3, PDHA1, PDHB, and SLC31A1) in pan-cancer. Results Cuproptosis-related genes (CRGs) were upregulated in most cancers tested. In KIRC, KIRP, LGG, MESO, and PCPG, most highly expressed CRGs predicted a better prognosis but poorer prognosis in patients with ACC, LIHC, and UCEC. Pathway analysis confirmed that cuproptosis regulators were associated with the metabolism-related pathways. The expression of MTF1, NLRP3, and SLC31A1 was positively related with ImmuneScore, StromalScore, and ESTIMATEScore in almost all types of tumor, whereas ATP7B, DLAT, DLD, LIAS, PDHA1, and PDHB were significantly negatively correlated with the scores. In addition, CRGs were significantly correlated with RNA stemness score, DNA stemness score, microsatellite instability, and tumor mutational burden. The expression of ATP7A, ATP7B, LIAS, and DLAT was significantly positively correlated with the drug sensitivity of Docetaxel. ATP7A, LIAS, and FDX1 were significantly negatively correlated with the drug sensitivity of UNC0638, XMD13−2, YM201636, and KIN001−260. Conclusions The altered genomic and clinical characteristics of cuproptosis regulators were comprehensively elucidated, providing a preliminary basis for understanding the functions of cuproptosis in pan-cancer. | Regulation, genomics, and clinical characteristics of cuproptosis regulators in pan-cancer
Cuproptosis, a copper-dependent controlled cell death, is a novel form of cell death that differs from known cell death mechanisms; however, its overall regulation in cancer remains elusive.
Multiple open-source bioinformatic platforms were used to comprehensively elucidate the expression levels, prognostic efficiency, potential biological functions, genomic and epigenetic characteristics, immune microenvironment, and drug sensitivity of cuproptosis regulators (ATP7A, ATP7B, DLAT, DLD, FDX1, GLS, LIAS, LIPT1, MTF1, NLRP3, PDHA1, PDHB, and SLC31A1) in pan-cancer.
Cuproptosis-related genes (CRGs) were upregulated in most cancers tested. In KIRC, KIRP, LGG, MESO, and PCPG, most highly expressed CRGs predicted a better prognosis but poorer prognosis in patients with ACC, LIHC, and UCEC. Pathway analysis confirmed that cuproptosis regulators were associated with the metabolism-related pathways. The expression of MTF1, NLRP3, and SLC31A1 was positively related with ImmuneScore, StromalScore, and ESTIMATEScore in almost all types of tumor, whereas ATP7B, DLAT, DLD, LIAS, PDHA1, and PDHB were significantly negatively correlated with the scores. In addition, CRGs were significantly correlated with RNA stemness score, DNA stemness score, microsatellite instability, and tumor mutational burden. The expression of ATP7A, ATP7B, LIAS, and DLAT was significantly positively correlated with the drug sensitivity of Docetaxel. ATP7A, LIAS, and FDX1 were significantly negatively correlated with the drug sensitivity of UNC0638, XMD13−2, YM201636, and KIN001−260.
The altered genomic and clinical characteristics of cuproptosis regulators were comprehensively elucidated, providing a preliminary basis for understanding the functions of cuproptosis in pan-cancer.
Cancer is a malignant disease with a pathological manifestation of abnormal cell proliferation accompanied by dysregulated cell death and a disturbed inflammatory response. In addition, it has high morbidity and mortality. Cell death is mainly classified as accidental cell death and regulated cell death (RCD) (1). RCD is mediated by a set of pathways that play an important role in development and immune responses (2). To date, researchers have identified several RCD mechanisms; among which, apoptosis, pyroptosis, necroptosis, and ferroptosis are the four most widely investigated forms in recent years (3–6). A recent study on copper-related death (cuproptosis) published in Science on 17 March 2022 (7) is the first to suggest that cuproptosis is a novel form of cell death that is copper-dependent, regulated, and different from other known mechanisms of RCD. Copper ions directly bind to lipoylated components in the tricarboxylic acid (TCA) cycle, leading to abnormal aggregation of lipoylated protein and the loss of iron–sulfur cluster proteins, thereby triggering proteotoxic stress responses and eventually mediating cell death. Copper levels are significantly altered in the serum and tumor tissue of patients in various cancers, including breast (8), thyroid (9), lung (10), colorectal (11, 12), oral cavity (13), prostate (14), and gallbladder (15) cancers. Growing evidence suggests that copper promotes angiogenesis, which is essential for tumor progression and metastasis development. This phenomenon is related to the fact that copper can activate angiogenesis-related factors, such as vascular endothelial growth factor, fibroblast growth factor 1, and angiopoietin (16, 17). Some studies have reported that high levels of copper can activate the function of Antioxidant 1 copper chaperone (ATOX1), increase the production of reactive oxygen species (ROS), and further enhance cell proliferation (18, 19). Immune escape of cancer cells is also an important mechanism of tumorigenesis, and studies have confirmed that copper in cancer cells can promote immune escape by overexpressing regulatory programmed death-ligand 1 (PD-L1) to protect the cells from tumor immune attack (20). Numerous studies have demonstrated that copper ion carriers increase the intracellular (especially mitochondrial) copper levels, which, in turn, increases ROS levels, eventually making cancer cells more susceptible to oxidative stress and leading to the development of cuproptosis (16, 17, 21). Therefore, copper metabolism plays a key role in the development of cancer and has emerged as a promising target for inhibiting cancer development. However, the complex relationship between cuproptosis-regulated genes and tumorigenesis requires further in-depth analysis. To gain insights into the mechanisms of cuproptosis-related genes (CRGs) in cancer, we comprehensively analyzed the transcriptomic, clinical, epigenomic, and immunological characteristics of 33 human cancers using The Cancer Genome Atlas Program (TCGA) data. Significant differences were found in mRNA expression, prognostic efficiency, epigenetic characteristics, and tumor immune microenvironment among CRGs, and they were enriched in multiple metabolic pathways, providing a rich resource for understanding cuproptosis biology.
On the basis of genome-wide CRISPR-Cas9 loss-of-function screening, Tsvetkov et al. found 10 regulatory genes specifically related to the cuproptosis metabolic pathway (7), including seven positively regulated genes [namely, ferredoxin 1 (FDX1), lipoic acid synthase (LIAS), lipoyltransferase-1 (LIPT1), dihydrolipoamide dehydrogenase (DLD), dihydrolipoamide S-acetyltransferase (DLAT), pyruvate dehydrogenase E1-alpha (PDHA1), and pyruvate dehydrogenase beta (PDHB)] and two negatively regulated genes [namely, metal transcription factor 1 (MTF) and glutaminase (GLS)]. Previous studies have also found that ATPase copper transporting alpha (ATP7A), ATPase copper transporting Beta (ATP7B), NLR family pyrin domain containing 3 (NLRP3), and solute carrier family 31 member 1 (SLC31A1) are closely associated with the cuproptosis metabolic process (12, 22–24). The Xena Functional Genomics Explorer database (Xena, https://xena.ucsc.edu/) (25) was used to extract normal tissues in the Genotype-Tissue Expression (GTEx) database and TCGA pan-cancer data, including transcription expression data, clinical data, immunological subtypes, and stemness scores {based on mRNA [RNA stemness scores (RNAss)] and DNA [DNA stemness scores (DNAss)] methylation}. TCGA pan-cancer data includes 33 cancer types, namely, adrenocortical carcinoma (ACC), bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), esophageal carcinoma (ESCA), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), kidney chromophobe (KICH), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), acute myeloid leukaemia (LAML), brain lower-grade glioma (LGG), lung squamous cell carcinoma (LUSC), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), mesothelioma (MESO), ovarian serous cystadenocarcinoma (OV), prostate adenocarcinoma (PRAD), pancreatic adenocarcinoma (PAAD), pheochromocytoma and paraganglioma (PCPG), rectum adenocarcinoma (READ), stomach adenocarcinoma (STAD), sarcoma (SARC), skin cutaneous melanoma (SKCM), testicular germ cell tumor (TGCT), thyroid carcinoma (THCA), thymoma (THYM), uterine corpus endometrial carcinoma (UCEC), uveal melanoma (UVM), and uterine carcinosarcoma (UCS). Tumor data without healthy tissue samples were removed, including MESO and UVM. Differential gene expression analysis between tumor and paracancerous tissues was performed using a linear mixed-effects model. The expression profile of CRGs in 1,062 cancer cell lines was obtained from Cancer Cell Line Encyclopedia (CCLE, http://www.broadinstitute.org/ccle/home) (26). The expression of CRGs in normal tissues in the GTEx database (27) was visualized on a heatmap.
The survival data of 33 TCGA cancer samples were extracted, including overall survival (OS), disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI) data. On the basis of a Cox proportional hazards regression model, the “survival” R package was used to investigate whether the expression of CRGs was associated with the survival of patients, and the “ggplot2” R package was used to draw a diagonal heatmap for visualizing results. Independent variables with a hazard ratio (HR) of >1 and <1 were referred to as risk and protective factors, respectively, with a threshold specified as a P-value of <0.05.
Gene Set Cancer Analysis (GSCALite, http://bioinfo.life.hust.edu.cn/GSCA/) is a user-friendly comprehensive cancer analysis database that integrates multigene, mutation, and drug sensitivity analyses spanning 33 cancer types in TCGA and Genomics of Drug Sensitivity in Cancer (GDSC) data (28). The tumor pathway activity module contains the activities of 10 cancer pathways, namely, apoptosis, cell cycle, DNA damage, epithelial–mesenchymal transition (EMT), hormone androgen receptor (AR), hormone estrogen receptor (ER), phosphatidylinositol-4,5-bisphosphate-3-kinase (PI3K)/protein kinase B (AKT), RAS/mitogen-activated protein kinase (MAPK), receptor tyrosine kinase (RTK), and tuberous sclerosis 1 protein (TSC)/mechanistic target of rapamycin (mTOR) pathways. The expression of CRGs was analyzed in relation to the activation or inhibition of the abovementioned oncogenic pathways. A protein–protein interaction (PPI) network of CRGs was constructed using the GeneMANIA platform (29) (http://genemania.org/), and functional enrichment analyses were performed to further understand the function of these genes in pan-cancer, including Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses.
cBioPortal platform (https://www.cbioportal.org/) can be used to study genetic alteration characteristics (30). cBioPortal precisely presented the details of all forms of the genetic alterations of structural variants, mutations, amplifications, deep deletions, and copy number alterations with the CRGs in pancancer by the OncoPrint module.
The copy number variation (CNV) module of the GSCALite platform contained CNV data of 11,495 samples from TCGA database, including homozygous and heterozygous deletion, diploid (none), and heterozygous and homozygous amplification, downloaded from TCGA database and processed by GISTIC2.0. The four CNV types of CRGs in pan-cancer were summarized using a pie chart, and Spearman correlation analysis was performed to analyze the correlation between the mRNA expression of CRGs and CNVs. Finally, survival differences between mutants and WT were analyzed using a Cox proportional hazards regression model and log-rank test.
The genome is considered to be affected by one conventional epigenetic alteration, DNA methylation. Using the methylation module of the GSCALite platform, the methylation data of 10,129 samples spanning 33 cancer types were extracted from the National Cancer Institute (NCI) Genomic Data Commons. Data on 14 cancer types were extracted, with paired healthy tissue data, and differences in methylation between cancerous and healthy tissue samples of these 14 cancer types (including BLCA, BRCA, COAD, ESCA, HNSC, KIRC, KIRP, LIHC, LUAD, LUSC, PAAD, PRAD, THCA, and UCEC) were analyzed using the Student’s t-test. Spearman correlation analysis was performed to analyze the correlation between the mRNA expression of CRGs and methylation. Finally, survival differences between the high- and low-methylation groups were compared using a Cox proportional hazards regression model and log-rank tests.
ESTIMATE is a tool used for predicting tumor purity (31). The expression data of CRGs were used to predict the number of infiltrating stromal/immune cells in tumor tissues and subsequently calculate the immune and stromal scores. Spearman correlation analysis was used to analyze the correlation between the genes and scores. The correlation between the expression of CRGs and six types of immune cells (B cells, CD4+ T cells, CD8+ T cells, neutrophils, macrophages, and dendritic cells) in 32 human cancers was analyzed using the Tumor IMmune Estimation Resource 2.0 website (Timer2.0, http://timer.cistrome.org/) (32) and visualized on heatmaps generated using the “ggplot2” R package. Thorsson et al. conducted an extensive immunogenomic analysis of 33 cancers in TCGA database and identified six immune subtypes, including C1 (wound healing), C2 (Interferon (IFN)-gamma dominant), C3 (inflammatory), C4 (lymphocyte depleted), C5 (immunologically quiet), and C6 (Transforming growth factor beta (TGF-b) dominant) (33). The pan-cancer immune subtypes were obtained from the Xena database, and the correlation between the expression of CRGs and the immune subtypes was assessed using ANOVA.
Tumor mutational burden (TMB) and microsatellite instability (MSI) are important biomarkers of the tumor microenvironment (TME) (34, 35). Spearman correlation analysis was used to analyze the correlation between the expression of CRGs and the abovementioned TME biomarkers. In addition, RNAss and DNAss can be used to assess the stem cell–like characteristics of tumors (36). Spearman correlation analysis was used to analyze the correlation between the expression of CRGs and RNAss/DNAss. Heatmaps were plotted using the “ggplot2” R package, and differences with a P-value of <0.05 were considered statistically significant.
The GDSC project collected 265 small molecules (37) and analyzed the correlation between the expression of CRGs and drug sensitivity according to the method used by Rees et al. (38). Spearman correlation was used to assess the correlation between gene expression and small-molecule drugs, in which a positive correlation indicated that samples with high gene expression were resistant to the drug. The NCI-60 cell line is the most widely used population of cancer cells for anti-cancer drug testing, and its gene expression and drug sensitivity data [including multiple Food and Drug Administration (FDA)–approved drugs and drug molecules in clinical trials] is available in the CellMiner database (https://discover.nci.nih.gov/cellminer/home.do) (39). Spearman correlation analysis was used to assess the correlation between the expression of CRGs and sensitivity to FDA-approved drugs.
Figure 1 illustrates the workflow of the study. Studies have reported that the expression of CRGs is significantly altered in various cancers. In this study, among the 33 TCGA cancer types, excluding cancer data without healthy tissue samples, 31 cancers were eventually included for differential expression analysis. Significant differential expression of almost all CRGs was observed among different cancer types (P < 0.05) and as upregulated in the majority of cancers tested ( Figure 2 ). Furthermore, the expression of CRGs was observed in various healthy tissues in the GTEx dataset. As shown in Supplementary Figure S1 the expression of DLD and FDX1 in the adrenal gland; ATP7A in the bladder; GLS in blood vessels; PDHA1 in the heart; SLC31A1 in the liver; DLAT and PDHB in muscle; NLRP3 in the spleen; and ATP7B, LIAS, LIPT1, and MTF1 in the testis was significantly upregulated. In addition, the expression of these genes in cancer cell lines was further analyzed using the CCLE dataset ( Supplementary Figure S2 ). The expression of ATP7A, ATP7B, DLAT, DLD, FDX1, GLS, LIAS, LIPT1, MTF1, NLRP3, PDHA1, PDHB, and SLC31A1 was higher in brain cancer, colon/colorectal cancer, rhabdoid, rhabdoid, bone cancer, kidney cancer, eye cancer, teratoma, thyroid cancer, leukemia, teratoma, rhabdoid, and fibroblast cell lines, respectively, than in other cancer cell lines.
A univariate Cox HR regression model was established in TCGA to analyzed four survival endpoints (OS, DSS, DFI, and PFI) to determine the prognostic value of CRGs. Survival analysis showed that most high CRG expression predicted a better prognosis in KIRC, KIRP, LGG, MESO, and PCPG, which both played a protective role. However, most higher expression levels of CRGs were associated with poorer survival in patients with ACC, LIHC, and UCEC ( Figure 3 , P < 0.05). Detailed results are provided in Table S1 . These results suggest that dysregulated expression of CRGs is associated with tumor prognosis.
We analyzed CRGs involved in well-known cancer-related signaling pathways (apoptosis, cell cycle, DNA damage response, EMT, hormone AR, hormone ER, PI3K/AKT, RAS/MAPK, RTK, and TSC/mTOR). In pan-cancer, CRGs were found to be closely associated with most well-known cancer-related pathways. Most CRGs tended to activate the apoptosis, cell cycle, PI3K-AKT, and RAS-MAPK pathways more than the inhibitory effects ( Figure 4A ). The set of genes closely regulating cuproptosis was obtained through the PPI regulatory network ( Figure 4B ) and subsequently subjected to functional enrichment analysis. In the GO biological process (BP) category, these genes were found to be mainly involved in the acetyl-CoA biosynthetic and metabolic processes (GO:0006086/GO:0006085/GO:0071616/GO:0006084/GO:0006637), the TCA cycle (GO: 0006099), and TCA metabolism (GO:0072350). In the molecular function (MF) category, these genes were found to be mainly involved in pyruvate dehydrogenase activity (GO:0004738/GO:0034603/GO:0034604). In the cellular component (CC) category, these genes were found to be mainly involved in pyruvate dehydrogenase complex (GO:0045254) and mitochondrial part (GO:0005967/GO:0005759/GO:0044429/GO:0005739) ( Figure 4C and Table S2 ). Furthermore, KEGG pathway analysis showed that these genes were enriched in multiple metabolic pathways, including lipoic acid metabolism, citrate cycle (TCA cycle), and glycolysis/gluconeogenesis. In addition, these genes were highly enriched in the HIF-1 signaling pathway, glucagon signaling pathway, and central carbon metabolism in cancer ( Figures 4D, E and Table S2 ). These results suggest that CRGs play an important role in human cancers through metabolism-related pathways.
To assess the mutation of CRGs in pan-cancer, we conducted an in-depth study using the cBioPortal database and found that the mutation frequency of ATP7A, ATP7B, DLAT, DLD, FDX1, GLS, LIAS, LIPT1, MTF1, NLRP3, PDHA1, PDHB, and SLC31A1 was 2.7%, 4%, 1.7%, 1.9%, 1.2%, 1.7%, 1%, 0.9%, 2.1%, 6%, 1.7%, 1.1.%, 0.7%, respectively. Amplification was the most common type of gene variation ( Figure 5 ). The CNV pie plot demonstrated that the two main types of CNVs were heterozygous amplification (Hete Amp) and heterozygous deletion (Hete Del) ( Figure 6A ), and the proportion of CNVs per gene in each cancer is shown in Table S3 . The correlation analysis showed that the mRNA expression of a vast majority of CRGs was positively correlated with CNVs (P < 0.05, Figure 6B ). Prognostic analysis revealed that the risk of death was higher for most CRGs in the CNV (mutant) group than in the WT group (P < 0.05, Figure 6C ). These results suggest that CNVs of CRGs mediate their aberrant expression, which may be closely related to the development and progression of cancer. DNA methylation, as one of the common epigenetic events, plays a key role in the diagnosis and treatment of tumors (40). In contrast to other CGRs, ATP7B, NLRP3, and ATP7A were distinctly hypomethylated in multiple cancers, such as BRCA, LUSC, and LUAD. PDHB was hypermethylated in COAD, KIRP, PAAD, HNSC, KIRC, LUSC, and BRCA (P < 0.05, Figure 7A ). The differential methylation levels of CRGs may be due to differences in expression patterns between tumor and normal tissues. To test this conjecture, we further assessed the correlation between DNA methylation levels and mRNA expression. ATP7B and NLRP3 showed a negative correlation in most cancers (P < 0.05, Figure 7B ). Prognostic analysis for OS, DSS, and PFS revealed that hypomethylation of DLAT, FDX1, MTF1, NLRP3, and PDHA1 was associated with low survival in most cancers, whereas hypermethylation of FDX1 and PDHB was associated with low survival in UVM(P < 0.05, Supplementary Figure S3 ).
Tumor tissue and the tumor immune microenvironment (TIME) are dependent on each other, and immune and inflammatory factors in the TIME play a key role in the immunotherapeutic response (41). In this study, TME analysis revealed that MTF1, NLRP3, and SLC31A1 were significantly positively correlated with TME scores (ImmuneScore, StromalScore, and ESTIMATEScore) in most human cancers, whereas ATP7B, DLAT, DLD, LIAS, PDHA1, and PDHB were significantly negatively correlated with the scores in most human cancers ( Figure 8 ; details are provided in Table S4 ). In addition, CRGs were significantly differentially expressed in different immune subtypes ( Supplementary Figure S4 ). ATP7B, DLAT, and PDHA1 were highly expressed in the C1 (wound healing) subtype; MTF1and SLC31A1 were highly expressed in the C2 (IFN-gamma dominant) subtype; ATP7A, FDX1, GLS, and LIPT1 were highly expressed in the C3 (inflammatory) subtype; DLD was highly expressed in the C4 (lymphocyte depleted) subtype, whereas LIAS and PDHB were highly expressed in the C5 (immunologically quiet) subtype; NLRP3 was highly expressed in the C6 (TGF-b dominant). Furthermore, a significant positive correlation was observed between the expression of CRGs and the infiltration of B cells, CD8+ T cells, macrophages, neutrophils, and dendritic cells across 32 cancer types ( Figure 9 ; details are provided in Table S5 ). These results suggest that dysregulated expression of the cuproptosis gene family may mediate disturbances in the TIME, which, in turn, promotes immunosuppression.
TMB and MSI in the TME are effective prognostic biomarkers and indicators of immunotherapeutic response in tumors (42–44). To examine the role of CRGs in anti-tumor immunity in the TME, we analyzed the correlation between the expression of these genes and TMB and MSI and found that CRG expression was significantly correlated with TMB and MSI in most tumors. For example, FDX1 expression was significantly positively correlated with TMB and MSI in HNSC, STAD, and UCEC but was significantly negatively correlated with TMB and MSI in LUAD (P < 0.05; Figure 10 , Table S6 ). These findings suggest the biological relevance of CGRs in PD1/PD-L1 therapy. In addition, the expression of CRGs was found to be associated with tumor stemness in pan-cancer ( Figure 11 , Table S6 ). The results showed that the expression of CRGs was significantly positively correlated with RNAss in a vast majority of tumors and that of ATP7A and NLRP3 was negatively correlated with RNAss. The expression of LIAS and LIPT1 was significantly positively correlated with DNAss, whereas that of FDX1, GLS, NLRP3, and PDHA1 was significantly negatively correlated with DNAss, with a large correlation coefficient.
The correlation between the expression of CRGs and drug sensitivity was analyzed in different human cancer cell lines using the GDSC and CellMiner™ databases ( Figure 12 ). Spearman correlation analysis showed that the half maximal inhibitory concentration (IC-50) values of Docetaxel were positively correlated with the expression of ATP7A, ATP7B, LIAS, and DLAT, whereas those of UNC0638, XMD13−2, YM201636, and KIN001−260 were negatively correlated with the expression of ATP7A, LIAS, and FDX1 ( Figure 12A ). Another database analysis showed that ATP7A expression was positively correlated with the drug sensitivity of ETHINYL ESTRADIOL and Estramustine but was negatively correlated with the drug sensitivity of Dasatinib and JNJ−42756493; GLS expression was positively correlated with the drug sensitivity of Ibrutinib but was negatively correlated with the drug sensitivity of TYROTHRICIN, Paclitaxel, and VINORELBINE ( Figure 12B ).
Elevated copper ion levels in cancer tissues can promote angiogenesis and immune escape, which, in turn, promotes tumor growth and metastasis (16, 20). In many clinical trials, different classes of copper ion carriers have been used for the treatment of cancer through mitochondrial oxidative stress and cuproptosis (44, 45). However, the field is in the early stages of development. Therefore, it is necessary to investigate the role of cuproptosis regulators in tumor development and to identify potential targets for clinical treatment. In this study, bioinformatic multiomic analyses provided a comprehensive understanding of the role of cuproptosis regulators in human cancers and propagated the search for new therapeutic targets. In this study, we investigated the differential expression of CRGs (ATP7A, ATP7B, DLAT, DLD, FDX1, GLS, LIAS, LIPT1, MTF1, NLRP3, PDHA1, PDHB, and SLC31A1) in cancer versus healthy tissues and healthy tissues versus cancer cell lines using the transcriptomic data derived from 33 different tumors, healthy tissues, and cancer cell lines in TCGA, GTEx, and CCLE datasets, respectively. Significant heterogeneity was found in the expression of these genes within tumors, between tumors and among cancer cell lines as well as between tumors in terms of prognosis, thus necessitating the study of each cuproptosis family member. A growing number of pan-cancer analyses have shown that genetic mutations are associated with tumorigenesis and progression (46). Our findings showed that the high frequency of copy number alterations in CRGs, mainly Hete Amp and Hete Del, and their significant positive correlation with gene expression accompanied by a higher risk of death suggest that copy number alterations may contribute to cancer development and progression by affecting gene expression. Moreover, compared with other CGRs, ATP7B, NLRP3, and ATP7A are significantly hypomethylated in various cancers, such as BRCA, LUSC, and LUAD. In addition, PDHB is hypermethylated in COAD, KIRP, PAAD, HNSC, KIRC, LUSC, and BRCA. Therefore, we hypothesized that hypomethylation induced the transcriptional activation of DLAT, FDX1, and PDHA1 genes, leading to cell death. Functional pathway analysis revealed that CRGs were closely associated with most well-known cancer-related pathways. It is now well established that copper intake is critical for the activity of MEK1 and MEK2 in the RAS/MAPK signaling pathway and for activating the phosphorylation of ERK1 and ERK2, and the activation of this pathway is a key factor in promoting tumor growth (47). Among the cuproptosis regulators analyzed in this study, most CRGs tended to activate the AS-MAPK pathways more than the inhibitory effects. In addition, functional enrichment analysis of the cuproptosis regulator PPI verified that cuproptosis regulators and their co-expressed genes are involved in various metabolic processes, especially TCA metabolism. Previous studies have found that copper, as a cofactor in the catalysis of most essential enzymes in the body, is frequently involved in energy production, oxygen transport, and cellular metabolism (48–50). Cancer cells have higher copper requirements than normal cells. Through the direct binding of copper to the fatty acylated component of TCA, some cancers abnormally express many lipoylated mitochondrial proteins and exhibit high-intensity respiration, which eventually leads to cell death (7). Therefore, cuproptosis regulators may constitute a network of interactions in cancer-related signaling pathways that promote the development of cuproptosis. The TME plays a crucial role in stimulating tumor cell heterogeneity, multidrug resistance, cancer progression, and metastasis (51). In the TME and immune cell infiltration correlation analysis, the expression of CRGs was found to be significantly correlated with immune scores and immune cell infiltration levels in some tumors, with a significant positive correlation between NLRP3 and TME in particular. Activation of NLRP3 inflammasome has been shown to affect inflammatory cell death, to mediate the secretion of pro-inflammatory cytokines, and to influence anti-tumor immunity (52). In a study, copper-treated mice exhibited ROS production and changes in the mitochondrial transmembrane potential, mainly leading to immunotoxicity in the form of reduced CD4+ T-cell populations and increased or proliferating CD4+ T-cell populations (53). In another study, copper chelators significantly increased the number of tumor-infiltrating natural killer cells, delayed tumor growth, and improved survival in mice by decreasing intracellular copper concentrations (20). In addition, copper chelators (CuNG) modulate the transition of tumor-associated macrophages from an immunosuppressed to a pro-immunogenic state (54). In this study, FDX1 expression was significantly positively correlated with immune scores and tumor immune cell infiltration in several human cancers. Moreover, for the regulation of cuproptosis, FDX1 is the key positive regulator of copper ion carrier–induced cell death, and high expression of FDX1 is accompanied by high immune cell infiltration, potentially leading to cell death through immunosuppressive effects. Overall, the interaction between the CRG family and immune cells in tumors may provide new perspectives for the development of more effective therapeutic strategies. During cancer progression, tumor cells gradually lose their differentiated phenotype and acquire progenitor and stem cell–like features, and stem cell indices are associated with active BPs of cancer stem cells and tumor dedifferentiation (55). In this study, the expression of CRGs was significantly correlated with DNAss and RNAss, with RNAss being significantly positively correlated with corresponding gene expression in most tumors. However, there were contradictory results of positive and negative correlations between the expression of CRGs and RNAss and DNAss of individual tumors. For example, in TCHA and THCA, FDX1 expression showed a significant positive correlation with RNAss and a significant negative correlation with DNAss. These contradictory results suggest that combining DNAss and RNAss can identify different tumor features or tumor cell populations characterized by different stemness. Another important finding of this study is the association between CRG expression and TMB and MSI in some cancer types. It is now well established that TMB and MSI can help to predict the response of patients to various drugs, particularly immune checkpoint inhibitors (35, 56). In addition, high TMB and MSI are associated with a good response to immune checkpoint inhibitors (57, 58). In this study, FDX1 expression was significantly positively correlated with TMB and MSI in HNSC, STAD, and UCEC, and DLAT expression was significantly positively correlated with TMB and MSI in READ, STAD, and UCEC, suggesting that these genes may be potential indicators of drug response. In addition, genes such as ATP7A, ATP7B, FDX1, GLS, and PDHA1 have been mined for identifying potential drug targets for drug sensitivity analysis. Tsvetkov et al. found that FDX1 is a direct target of elesclomol and both act in correlation, with increased ROS production owing to increased copper uptake, eventually leading to copper-dependent cell death in cancer cells (59). FDX1 has been reported to enhance the copper-dependent cell death induced by elesclomol, providing a new idea to improve the efficacy of cancer-targeted drugs (59). Therefore, detecting the expression of CRGs in patients with cancer is relevant for guiding the selection of clinical drugs. However, further studies are required to verify these results and clarify the potential mechanisms underlying drug regulation of CRGs. Although this is the first study to multidimensionally analyzed CRGs across multiple cancer types, it has some limitations. First, all results are based on public databases and have not been validated using other independent databases. Second, the underlying mechanisms behind the bioinformatic analysis have not been explored through molecular and animal experiments. Therefore, the specific biological role and mechanism of CRGs warrant further experimental validation. The study provides a comprehensive analysis of CRGs in pan-cancer. Upregulation of CRG expression was observed in most cancers compared with normal tissues. Survival analysis confirmed that most highly expressed CRGs had a better prognosis in KIRC, KIRP, LGG, MESO, and PCPG and a worse prognosis in patients with ACC, LIHC, and UCEC. The pathway results suggest that CRGs are mainly involved in tumor metabolism-related signaling pathways. In addition, CRGs correlated with the level of immune cell infiltration, TMB, MSI, and tumor stemness score, with NLRP3 being more strongly correlated. In conclusion, these findings may provide new insights into CRGs as potential therapeutic targets in pan-cancer.
Publicly available datasets were analyzed in this study. The datasets presented in this study can be found in online repositories. The name and login number of the repository/repository are as follows: TCGA and GTEx form Xena (https://xena.ucsc.edu/), CCLE (http://www.broadinstitute.org/ccle/home), GSCA (http://bioinfo.life.hust.edu.cn/GSCA/), Timer2.0 (http://timer.cistrome.org/), GDSC (https://www.cancerrxgene.org/), TISIDB (http://cis.hku.hk/TISIDB/) and CellMiner (https://discover.nci.nih.gov/cellminer/home.do).
CZ and YZ performed data analysis work and aided in writing the manuscript. CZ and XH designed the study and assisted in writing the manuscript. CL edited the manuscript. All authors read and approved the final manuscript.
This work was supported by grants from the Foshan Transnatural Cavity Surgery Development and Innovation Engineering Technology Research Center (2020001003507) and 2020 Foshan City self-financing science and technology projects (2020001005656).
We thank Bullet Edits Limited for the linguistic editing and proofreading of the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647019 | Thu Huyen Nguyen,Huy Hoang Dao,Chau Minh Duong,Xuan-Hung Nguyen,Diem Huong Hoang,Xuan-Hai Do,Trung Quang Truong,Tu Dac Nguyen,Liem Thanh Nguyen,Uyen Thi Trang Than | Cytokine-primed umbilical cord mesenchymal stem cells enhanced therapeutic effects of extracellular vesicles on osteoarthritic chondrocytes | 27-10-2022 | osteoarthritis,chondrocytes,mesenchymal stem cells,extracellular vesicles,cytokines,microRNA | In recent years, extracellular vesicles (EVs) secreted by mesenchymal stem cells (MSCs) have emerged as a potential cell-free therapy against osteoarthritis (OA). Thus, we investigated the therapeutic effects of EVs released by cytokine-primed umbilical cord-derived MSCs (UCMSCs) on osteoarthritic chondrocyte physiology. Priming UCMSCs individually with transforming growth factor beta (TGFβ), interferon alpha (IFNα), or tumor necrosis factor alpha (TNFα) significantly reduced the sorting of miR-181b-3p but not miR-320a-3p; two negative regulators of chondrocyte regeneration, into EVs. However, the EV treatment did not show any significant effect on chondrocyte proliferation. Meanwhile, EVs from both non-priming and cytokine-primed UCMSCs induced migration at later time points of measurement. Moreover, TGFβ-primed UCMSCs secreted EVs that could upregulate the expression of chondrogenesis markers (COL2 and ACAN) and downregulate fibrotic markers (COL1 and RUNX2) in chondrocytes. Hence, priming UCMSCs with cytokines can deliver selective therapeutic effects of EV treatment in OA and chondrocyte-related disorders. | Cytokine-primed umbilical cord mesenchymal stem cells enhanced therapeutic effects of extracellular vesicles on osteoarthritic chondrocytes
In recent years, extracellular vesicles (EVs) secreted by mesenchymal stem cells (MSCs) have emerged as a potential cell-free therapy against osteoarthritis (OA). Thus, we investigated the therapeutic effects of EVs released by cytokine-primed umbilical cord-derived MSCs (UCMSCs) on osteoarthritic chondrocyte physiology. Priming UCMSCs individually with transforming growth factor beta (TGFβ), interferon alpha (IFNα), or tumor necrosis factor alpha (TNFα) significantly reduced the sorting of miR-181b-3p but not miR-320a-3p; two negative regulators of chondrocyte regeneration, into EVs. However, the EV treatment did not show any significant effect on chondrocyte proliferation. Meanwhile, EVs from both non-priming and cytokine-primed UCMSCs induced migration at later time points of measurement. Moreover, TGFβ-primed UCMSCs secreted EVs that could upregulate the expression of chondrogenesis markers (COL2 and ACAN) and downregulate fibrotic markers (COL1 and RUNX2) in chondrocytes. Hence, priming UCMSCs with cytokines can deliver selective therapeutic effects of EV treatment in OA and chondrocyte-related disorders.
Extracellular vesicles (EVs) are nano-sized, lipid membrane-enclosed particles that modulate the physiological conditions of the recipient cells (1). By effectively delivering a wide range of bioactive molecules involved in critical signaling pathways associated with apoptosis, proliferation, migration, extracellular matrix (ECM) synthesis, cartilage regeneration, and inflammation management, EVs have been studied for their therapeutic effects on several cartilage-related diseases (2). Recently, multiple approaches have been employed to enhance therapeutic effect, and targetable EV delivery, including engineering secreted cells, loading therapeutic molecules into naturally secreted vesicles (3), and conjugating the vesicles with targeting ligands (4). Additionally, the therapeutic cargo of EVs secreted by mesenchymal stem cells (MSCs) varies depending on MSC tissue sources (5). Since MSCs are very sensitive to environmental conditions, priming these cells with cytokines as supplements in the culture media can influence bioactive molecules packed in the derived EVs, thereby affecting the biological activities of the vesicles (6, 7). In this study, we primed MSCs originated from the umbilical cord (UCMSCs) with anti-inflammatory cytokines (transforming growth factor beta - TGFβ and interferon alpha - IFNα) and inflammatory cytokines (including tumor necrosis factor alpha - TNFα), which are linked with osteoarthritis (OA) pathogenesis. In healthy cartilage, TGFβ stimulates chondrocyte proliferation while suppressing chondrocyte hypertrophy and maturation, as well as promoting chondrocytes to synthesize ECM components (8). Additionally, the inhibition of TGFβ signaling leads to chondrocyte terminal differentiation and the early onset of OA (9). Another stimulant in the anti-inflammatory, IFNα, plays a vital role in autoimmunity and inflammation and could effectively protect against antigen-induced arthritis by inhibiting pro-inflammatory cytokine (interleukin 1β (IL-1β), IL-6, IL-17, TNF, IL-12, and IFNγ) production while inducing TGFβ synthesis (10). Moreover, injection of IFNα into the synovial fluid promotes the generation of functional antagonists such as interleukin 1 receptor antagonist (IL-1Ra), soluble tumor necrosis factor receptors (sTNFR), and osteoprotegerin (OPG) for known OA-inducing factors of IL-1, TNF, and osteoprotegerin ligand (11). However, direct administration of these two cytokines to patients frequently harms the nearby tissues, such as a synovial membrane or subchondral bone, as well as general health, including headache, malaise, fever, and even depression (12). The inflammatory cytokine TNFα which is one crucial catabolic factor for cartilage, promotes synovial fibroblasts to release Matrix metalloproteinase (MMPs), resulting in cartilage destruction during OA progression (13). This cytokine is also able to signal chondrocyte apoptosis leading to a more severe OA phenotype (13). Moreover, priming UCMCS with cytokines enhances the anti-inflammatory and immunomodulatory potential of the secreted EVs (14–16). TNFα stimulation was shown to induce the expression of immunosuppressive factors in the parental MSCs, which produced exosomes that can modulate M2/M1 macrophage differentiation (14). The molecular content changes in EV derived from cytokine-stimulated MSCs can interfere with inflammation via PGE2/COX10 mechanism (15). Although the immunomodulatory effect of EVs from cytokines-primed cells in the context of OA has been reported before (14, 17), no publication was found indicating their influence on chondrocytes. In OA, several miRNAs found in EVs have been demonstrated to regulate key signaling pathways involved in ECM maintenance, chondrocyte proliferation, migration, apoptosis, and inflammation (2). Thus, modulating miRNA composition might directly influence the EV therapeutic effect. In this study, we focused on two candidate miRNAs, miR-320a-3p and miR-181b-3p, which involve in cartilage homeostasis. Previous studies showed that miR-320a played essential roles in the secretion of matrix degradation factors (18) and chondrocyte proliferation (19) in OA models. Although not many studies focused on miR-181b-3p, it was described to inhibit proliferation as well as promote apoptosis of chondrocytes in OA (20). As described, different EV sub-populations carry a different set of bioactive molecules (21), for instance, miRNAs content; thus, we assumed that they would have a distinct impact on chondrocytes. We further hypothesized that priming UCMSCs with the cytokines could alter the miR-181b-3p and miR-320a-3p levels in the secreted EVs, thereby modulating the effect of EVs in chondrocytes proliferation, migration, and their markers.
We examined the morphology and cell surface markers of UCMSCs at passage 5 (P5), either non-priming or cytokine priming. We observed a typical UCMSC morphology with a spindle shape in both control (EV-depleted) and cytokine-primed (TGFβ, IFNα, TNFα) culture conditions ( Figures 1A1-A4 ). Additionally, all cultured UCMSCs expressed MSC positive markers of CD90, CD105, and CD73 (> 95%). Meanwhile, MSC negative markers of CD45, CD34, CD11b, CD19, and HLA-DR were detected with low percentages (< 2%) ( Figure 1B ). Hence, cytokine priming did not alter the typical morphology and surface markers of UCMSCs. EVs isolated from UCMSC culture medium were subjected to morphology analysis by transmission electron microscope (TEM). Three EV sub-populations including apoptotic bodies (ABs), microvesicles (MVs), and exosomes (EXs), were observed with distinguished shapes and sizes ( Figures 1C1 - C3 ). ABs showed variable shapes with a diameter scale of approximately 500 nm to 2000 nm, and they are packed within a rough membrane ( Figure 1C1 ). MVs had variable membrane-bound morphologies with uneven surfaces and diameters ranging from 100 nm to 1 µm ( Figure 1C2 ). EXs exhibited a typical cup-shaped morphology with their size ranging from 40 to 200 nm ( Figure 1C3 ). Additionally, the isolated EVs expressed standard EV protein markers (CD9 and CD63). As a control indicator, all three EV populations strongly expressed the internal reference protein of GAPDH. For general EV marker expression, CD9 was present abundantly in MVs and EXs and lightly in ABs. EXs also expressed a typical exosomal marker of CD63, which was absent in MVs and ABs ( Figure 1D ). The morphology and protein marker analysis confirmed the identity of three separated EV populations from UCMSC conditioned media.
We measured the expression of selected miRNAs associated with OA pathogenesis present in UCMSCs and three secreted EV populations (ABs, MVs, and EXs) from normal and cytokine-primed conditions. Using qRT-PCR, we quantified the levels of two candidate miRNAs: miR-181b-3p and miR-320a-3p. Generally, both miR-181b-3p and miR-320a-3p were detected in all UCMSCs and isolated EV sub-populations of ABs, MVs, and EXs from different culture conditions ( Figure 2 and Supplementary Figure 1 ). In UCMSCs, cytokine treatments acted differentially on the expression of two candidate miRNAs. Particularly, TGFβ and TNFα significantly induced the levels of miR-320a-3p present in UCMSCs, indicated by lower delta Ct values ( Supplementary Figure 1A ). Besides, no significant impact was detected on the expression of miR-181b-3p ( Supplementary Figure 1B ). In contrast, cytokine-priming significantly modulated the levels of miR-181b-3p while producing a little impact on the levels of miR-320-3p packed into EVs. Cytokine treatment suppressed the selective sorting of miR-181b-3p in all three EV sub-populations compared to the non-priming group, indicated by higher delta Ct values when normalizing to secreted cells, UCMSCs ( Figure 2B ). Comparing the effects among different cytokines in each EV sub-populations, IFNα and TNFα cytokine treatments further limited the miR-181b-3p content in ABs and MVs ( Figure 2B ). Indeed, we detected a greater relative expressionof miR-181b-3p in TGFβ-ABs compared to IFNα-ABs (p = 0.0359) and TNFα-ABs (p = 0.0101), as well as in TGFβ-MVs compared to TNFα-MVs (p = 0.028). miR-181b-3p also expressed stronger in MVs from IFNα-primed UCMSCs compared to TNFα-primed ones (p = 0.038) ( Figure 2B ). However, in the EX population, no significant difference in miR-181b-3p inhibiting effects was observed among three different cytokine priming conditions ( Figure 2B ). Meanwhile, the amount of miR-320a-3p packed in ABs, MVs, and EXs was mostly stable in all non-priming and cytokine-priming cultures. Only a small change of miR-320a-3p content in the EX population was detected under the effect of TGFβ and TNFα, where this miRNA was relatively expressed higher in CT-EXs and TNFα-EXs compared to TGFβ-EXs (p = 0.0299 and p = 0.0031, respectively) ( Figure 2A ). Taken together, cytokine-primed UCMSCs selectively reduced the sorting of miR-181b-3p into EVs, whereas there were no effects on the miR-320a-3p.
Human chondrocytes were isolated from the knee articular cartilage tissue digested with collagenase and were cultured in the DMEM/F12. As shown in Figure 3A , cells at P1 exhibited flattened and polygonal shape which is a typical morphology of chondrocytes. Additionally, isolated cells were positive with Alcian Blue staining dye, which is specific for chondrocyte cells and appears blue due to proteoglycans secretion ( Figure 3B ).
We performed an MTT assay on human chondrocytes to examine the effect of EVs derived from cytokine-priming UCMSCs on chondrocyte proliferation. In general, EVs generated from UCMSCs, either priming with cytokines or not, did not have any statistically significant effect on chondrocyte proliferation compared to EV-depleted media (No-EV) and among treatment groups ( Figure 4 ).
We performed the wound scratch assay to access the capacity of EVs from cytokine-primed UCMSCs in regulating chondrocyte migration. In general, EVs from either non-priming or cytokine-primed UCMSCs significantly promoted chondrocyte migration starting from the 44-hour time point (later experimental time points) (CT-ABs, p = 0.0445 at 68 hours; and CT-MVs, p = 0.0105 at 44 hours; when compared to EV-depleted media) ( Figure 5 and Supplementary Figure 2 ). Among analyzed cytokines, EVs derived from TGFβ-primed UCMSCs significantly stimulated cell migration stronger than EV-depleted media at multiple time points. For details, TGFβ-ABs had higher migration induction at 52 hours (p = 0.0169) ( Figure 5A ). TGFβ-MVs enhanced migration at 48 hours (p = 0.0329), 52 hours (p = 0.0108), and 68 hours (p = 0.0129) ( Figure 5B ). Additionally, TGFβ-MVs exhibited a greater induction on cell migration than CT-MVs at 52 hours (p = 0.0215) and at 68 hours (p = 0.0424) and better than IFNα-MVs at 68 hours (p = 0.0426) ( Figure 5B ). At 68 hours, TGFβ-EXs significantly promoted migration stronger than No-EV (p = 0.0351) ( Figure 5C ). Regarding IFNα cytokine priming, only IFNα-EXs induced cell migration faster than No-EV, but neither IFNα-ABs nor IFNα-MVs, at 44 hours (p = 0.0319), 52 hours (p = 0.0376), and 68 hours (p = 0.0338) ( Figures 5A, B ). Especially, IFNα-ABs and IFNα-MVs showed suppression of chondrocyte migration at the 20-hour time point (compared to No-EV; p = 0.0259 and p = 0.0190, respectively) ( Figures 5A, B ). For EVs secreted from TNFα-primed UCMSCs, TNFα-MVs stimulated migration more effectively than No-EV at 44 hours (p = 0.0456), whereas TNFα-EXs promoted efficiently at 44 hours (p = 0.0492) and 48 hours (p = 0.0128) ( Figures 5B, C ).
To investigate the molecular alterations of chondrocytes in different EV-treated culture conditions, we isolated total RNA from cells after one-week culture under EV treatment and subjected them to qRT-PCR. The relative expression level of chondrocyte mRNAs, including normal chondrocyte markers of Collagen type II (COL2A1), Cartilage oligomeric matrix protein (COMP), Aggrecan (ACAN), and hypertrophic chondrocyte markers of Collagen type I (COL1A1), Runt-related transcription factor 2 (RUNX2), were calculated and represented as fold change. In general, the treatment with either normal EVs or EVs associated with cytokine priming acted differentially on the expression of chondrocyte markers. We observed the highest expression of COL2A1 in chondrocytes treated with TGFβ-MVs in all experimental groups ( Figure 6A ). Notably, non-priming EVs greatly enhanced the expression of COMP in chondrocytes, much stronger than any studied groups ( Figure 6B ). When analyzing the expression of ACAN, we observed that treatment with TGFβ-MVs significantly upregulated ACAN mRNA expression by chondrocytes compared to EV-depleted media (p = 0.0156) ( Figure 6C ). Regarding hypertrophic markers, treating chondrocytes with EVs from both cytokine-primed and non-priming groups suppressed COL1A1. The downregulation of COL1A1 was indicated by significantly lower expression levels in chondrocytes treated with CT-MVs and CT-EXs (p = 0.0195, and p = 0.0185, respectively); all three TGFβ-EV sub-populations (TGFβ-ABs, p = 0.0135; TGFβ-MVs, p = 0.0373; and TGFβ-EXs, p = 0.0282), IFNα-EXs (p = 0.0411), all three TNFα-EV sub-populations (TNFα-ABs, p = 0.0019, TNFα-MVs, p = 0.0003; and TNFα-EXs, p = 0.0054) ( Figure 6D ). Interestingly, the inhibition of COL1A1 mRNA was further emphasized in the chondrocytes treated with TNFα-EVs, showed by a reduced expression of COL1A1 by chondrocytes treated with TNFα-MVs compared to IFNα-MVs (p = 0.0310) ( Figure 6D ). In contrast with COL1A1, the expression of RUNX2 was upregulated by chondrocyte treatment with EVs in general. Indeed, compared to cells cultured in EV-depleted media, RUNX2 was expressed higher in chondrocytes treated with non-priming (CT) EVs (CT-ABs, p = 0.0498; CT-MVs, p = 0.0128; and CT-EXs, p = 0.0012), TGFβ-MVs (p = 0.01) ( Figure 6E ). However, priming cells with cytokines seemed to diminish this undesirable effect with lower expression of RUNX2 in chondrocytes treated with MVs and EXs from IFNα- and TNFα-primed cells compared to chondrocytes treated with CT-EVs and TGFβ-EVs ( Figure 6E ). Taken together, treatment of chondrocytes with cytokine-primed EVs partially rescued the chondrocytes from hypertrophic phenotype and established the primary, normal physical state of the cells.
In recent years, several techniques have been developed to optimize the therapeutic efficacy of EVs. Evidently, adding cytokines, such as IFNγ, TNFα, and IL1β, to the conventional MSC culture media affected the contents and biological activities of the derived EVs associated with OA (17, 23). Therefore, we investigated the influence of EVs derived from MSCs primed with anti-inflammatory (TGFβ and IFNα) and inflammatory (TNFα) cytokines on osteoarthritic chondrocytes. We found that cytokine priming did not affect the typical morphology and markers of UCMSCs. Additionally, the secreted EVs displayed distinguished sizes, morphologies, as well as surface markers (CD9 and CD63), which were in accordance with ISEV guidelines (24). These characteristics are also described in the previous studies of EVs from cytokine-primed UCMSCs (25, 26). This information allowed us to ensure the normal EV identity and further study the molecular profile and therapeutic effects of EVs under cytokine stimulations. Furthermore, stimulation of secreted cells with cytokines can also increase the amount of EVs produced by UCMSCs (14), which can potentially enrich therapeutic efficacy. Hence, the assessment of EV production from cytokine-stimulated UCMSCs should be considered in our future investigation. As mentioned, the treatment of UCMSCs with various stimuli could affect the biological contents of EVs, including miRNAs, which have been reported for their potential roles in OA treatment (27, 28). Especially, it is emphasized that cytokine treatment can result in EVs with rich RNA profiles for inflammatory control (15), which can further reverse OA condition. This study reported the detection of miR-320a-3p and miR-181b-3p, which are involved in healthy cartilage maintenance and OA pathogenesis (18–20) in MSCs and three EV sub-populations released by UCMSCs. Our result indicates that miR-320a-3p was expressed higher in non-priming UCMSCs, while the expression level of miR-181b-3p was similar among groups. However, cytokines treatment diminished the packaging of miR-181b into EVs, shown by a significantly low relative expression of this EV miRNA associated with cytokine priming when normalized to the levels in UCMSCs. In literature, miR-181b promoted the NF-κB pathway, which leads to cartilage destruction and synovium membrane degradation (29–31). Blocking miR-181b activity reduced MMP13 expression but increased COL2 expression in articular chondrocytes (32). The attenuation of miR-181b activity can indirectly signal the FPR2- formyl peptide receptor and induce anti-inflammatory effects (33, 34). Thus, the reduction of miR-181b observed in EVs originating from cytokines-primed cells can be a positive marker for re-adjusting the appropriate EV components to produce more direct effects in cartilage regeneration. This exciting information requires further studies to validate whether two cytokines, IFNα and TNFα, could be the appropriate stimulus to enhance the therapeutic efficacy of UCMSC-EVs for OA treatment. On the other hand, the level of miR-320a-3p remained stable across experimental EV treatments. Previous studies showed pieces of diverse evidence of miR-320a function in cartilage homeostasis (18, 19, 35). Peng et al. (19) also demonstrated the protective effects of miR-320a over cartilage degeneration by negatively regulating BMI-1 (19). However, miR-320a has also been shown as a potential OA marker as this miRNA promoted OA-induced matrix breakdown via the NF-κB pathway and interfered with osteoblast reformation (18, 35, 36). Thus, a future study is required to evaluate the roles of miR-320a-3p in OA pathogenesis and examine an alternative approach to adjusting this miRNA content in UCMSC-derived EVs. Next, to examine our EVs’ bioactivity in vitro, we isolated human primary chondrocytes from articular cartilage tissues obtained from a patient suffering from a knee injury and performed proliferation, migration and mRNA markers analysis assays. Our isolated cells showed typical chondrocyte morphology and were positive with specific staining dye for proteoglycan. For functional analysis, in general, all EVs from either non-priming or cytokines-primed UCMSCs at the dose of 10 μg/mL did not promote chondrocyte proliferation significantly. This result may be due to an insufficient dose of EVs might be the issue, as higher doses (20, 40, 80 µg/mL) of BMMSC-EXs have been shown to increase the proliferation rate of chondrocytes (37). Additionally, the outcomes may be due to chondrocytes obtained from patients with knee injury reported herein instead of healthy chondrocyte cell lines. Hence, further experiments with chondrocytes induced with OA characteristics and higher EV dosage will be conducted to examine these possibilities. Notably, the miRNA distribution in EVs is an essential factor that might affect chondrocyte proliferation and migration; however, the regulation of miR-181b on these two biological processes remains unclear. A member of the miR-181 family, miR-181a, exerted adverse effects on chondrocyte proliferation by upregulating the expression of caspase-3, PARP, MMP-2, and MMP-9 to induce apoptosis and cartilage destruction (20). Thus, it is predicted that miR-181b can inhibit chondrocyte proliferation, and suppressing miR-181b expression can restore this ability. However, in this current study, the reduction in miR-181b might not contribute to proliferation results observed here, or other factors have surpassed its influence. Contrary to cell proliferation, EVs from cytokines-primed UCMSCs expressed a higher capacity to promote chondrocyte migration compared to chondrocytes cultured in EV-depleted media, with the most significant effect belonging to TGFβ-EVs but only in later time points. In the previous study, TGFβ was shown to promote the PI3K-Akt signaling pathway, which was demonstrated to induce chondrocyte migration in a rat model (38, 39). Additionally, TGFβ stimulation can also regulate the integrin signaling pathway involving changes in integrin-ECM binding and the activation of FAK, which are critical factors in cell migration (40, 41). It is noted that all results obtained herein were in the comparison with chondrocytes cultured in EV-depleted media (DMEM/F12 supplemented with 5% EV-depleted FBS) but not as in most studies used PBS or chamber consisting of low serum media (upper) and PBS (lower) as the control group (42–44). Indeed, long-term cell storage with PBS increases cell death and thus cannot access cell functionality efficiently. Those factors may be reasons for the differences observed in this study compared to others. In this study, we investigated the alteration in mRNA levels of chondrocyte markers under the treatment of EVs at high passage culture. The later culture passage exhibited an increase in COL1 and a decrease in COL2. However, higher expression of healthy chondrocyte markers, including COMP and ACAN, was also detected, which supports cartilage regeneration and ECM synthesis at the later stage. Meanwhile, the expression of hypertrophic markers such as RUNX2 diminished. Besides, we observed that EVs from cytokines-primed UCMSCs downregulated the expression of COL1 and RUNX2 and upregulated COL2 and ACAN expression, but this effect was not consistent among EV populations. MiR320a was previously linked with low expression of hypertrophic marker RUNX2 (19). However, a stable level of miR-320a-3p in most of the isolated EVs from both non-priming and cytokine-primed UCMSCs hinders us from revealing the association between EV contents and chondrocyte markers. Notably, the increase in COMP expression was much more substantial in chondrocytes treated with EVs derived from non-priming UCMSCs. This means that EVs contribute to chondrocyte malfunctions or the recovery of damaged chondrocytes. However, further experiments, especially on miRNAs and target mRNAs, should be investigated to understand the mechanism of the effect of EV contents on chondrocyte mRNA expression. In conclusion, cytokines influenced the miRNA composition of UCMSCs-derived EVs and their effects on chondrocyte physiology regarding cell proliferation and migration, as well as chondrocyte markers. However, it is noted that the results presented here are preliminary data that require more investigations on other miRNAs/proteins found in EVs in addition to the target genes and signaling pathways affecting the chondrocyte bioactivities. Additionally, for future perspectives, studies should be performed to examine the roles of different cytokines on UCMSC-derived EVs and their cargos in other aspects of OA, such as chondrocyte apoptosis and inflammation.
Ethical approval for collecting and using human MSCs from the umbilical cord and human chondrocytes from articular cartilage was issued by the Vinmec International General Hospital Joint Stock Company’s ethics committee (Ethical approval number: 311/2018/QĐ-VMEC). The umbilical cord tissues were collected from three healthy donors aged 20 to 40, and human cartilage tissues were acquired from three donors with knee arthroplasty. Donors signed written informed consent before donating their samples.
UCMSCs were isolated from the umbilical cord following what was described in our previous study and stored for further experiments (45). UCMSCs at passage two (P2) were thawed and seeded at a density of 5,000 cells/cm2 in DMEM/F12 (Gibco, Massachusetts, USA) with 10% (v/v) fetal bovine serum (FBS). Cells were incubated in 37°C/5% CO2 condition and subcultured with the same density until passage 5 (P5). The cells at P5 were cultured with EV-depleted media for three days prior to cytokine treatments (DMEM/F12 supplemented with 10% EV depleted-FBS, in which FBS was centrifuged at 120,000 × g for 18 hours at 4°C to eliminate EVs). Cells were maintained in EV-depleted media before exposing to cytokines individually for 48 hours with the following concentrations: 10 ng/mL TGFβ, 20 ng/mL IFNα, or 20 ng/mL TNFα. The conditioned media were harvested when cells reached 95% confluency for EV isolation (cell culture media were not renewed throughout incubation). After conditioned media collection, UCMSCs were characterized with Human MSC Analysis Kit (BD Biosciences) following the manufacturer’s protocol, and flow cytometry data were analyzed with Navios Software 3.2.
The conditioned media was centrifuged at 300 × g for 10 minutes at 4°C to remove cell debris. Sequential centrifugation steps were performed to separate three EV populations as follows: 2,000 × g for 20 minutes at 4°C to collect apoptotic bodies (ABs), 16,500 × g for 30 minutes at 4°C to pellet microvesicles (MVs), and 100,000 × g for 90 minutes at 4°C for isolation of exosomes (EXs) (Optima XPN-100 Ultracentrifuge, Beckman Coulter, California, USA). EV pellets were resuspended in DMEM/F12 or PBS and stored at − 80°C for further usage.
Protein extraction and western blot were performed as described previously (45). Total EV protein concentrations were determined by Pierce™ BCA Protein Assay Kit (Thermo Scientific, Massachusetts, USA) and as equivalent to an optical density (OD) measured at 562 nm (SpectraMax M3, Molecular Devices, California, USA). Then, 15 µg of EV proteins were electrophoretically separated by 4 – 12% NuPAGE gels (Invitrogen, Massachusetts, USA) and probed with primary antibodies (Abcam, Cambridge, UK) against GAPDH, CD9, and CD63 overnight at 4°C, followed by the incubation with goat anti-Rabbit IgG secondary antibody (Invitrogen, Massachusetts, USA). Antibody binding was stained with ECL substrate and visualized on ImageQuant LAS 500 (GE Healthcare Life Sciences, Illinois, USA).
EV samples were fixed and stained following the protocol described in our previous study (45). Imaging was performed using a JEOL 1100 Transmission Electron Microscope (JEOL Ltd., Tokyo, Japan) at 80 kV at the National Institute of Hygiene and Epidemiology (NIHE).
Human cartilage tissues were collected by the surgical doctors, stored in saline water at 4°C, and transferred to the laboratory. Before processing, the tissue was washed once with ethanol 70%, twice with PBS, and once with DMEM/F12; each solution was supplemented with 1% Pen/Strep (Thermo Fisher Scientific, USA) to ensure sterile and eliminate contaminants. The tissue was minced and digested in Hanks’ Balanced Salt Solution (HBSS) (Thermo Fisher Scientific, USA) 0.2% collagenase type I 10000 U/mL solution (Gibco, Massachusetts, USA) (10 mL for every 1 gram of tissue) for 20 hours at 37°C. Cell culture media (DMEM/F12 supplemented with 10% FBS (v/v)) was added in a volume ratio of 1: 1 with HBSS. The harvested pellets were resuspended in DMEM/F12 supplemented with 1% Pen/Strep and 10% FBS (v/v), then seeded into a T25 cell culture flask and incubated at 37°C and 5% CO2. The media were replaced by every three days during the cultures. After reaching 80% confluency, the cells were either stored or subcultured at the density of 10,000 cells/cm2 to the next passage. The images of chondrocytes were captured under Eclipse Ti-S Inverted Microscope (Nikon Instruments, Japan), and cells at P0 were processed to form the colony and stained with Alcian Blue to confirm cell type.
Total RNA was extracted using Trizol™ reagent (Thermo Scientific, Massachusetts, USA) with a ratio of 9: 1 Trizol versus cell/particle suspension. The lysis mixture was added with MgCl2 and chloroform and incubated at RT. The aqueous phase was collected and incubated in with isopropanol overnight at -20°C. Total RNA was then pelleted with centrifugation and then washed twice with RNase-free 75% ethanol before air-drying and resuspending in RNase-free water (volume based on pellet size).
Total RNA with sufficient quality was subjected to qRT-PCR to confirm the presence of EV miRNAs and chondrocyte mRNAs. For EV miRNA analysis, extracted RNAs were used as templates to prepare cDNA using the miScript II RT kit (Qiagen, Hilden, Germany), following the manufacturer’s instructions. Then, cDNA-containing mixtures (10 μL) were subjected to qPCR using the miScript SYBR Green PCR kit (Qiagen, Hilden, Germany) and two specific primers, miScript Primer Assay 10X (Qiagen, Hilden, Germany) designed to target miR-320a-3p and miR-181b-3p. The incubation was performed on Applied Biosystems 7500 Block (Applied Biosystems, Massachusetts, USA). The relative expression of miRNAs in UCMSCs was normalized to reference gene RNU6B (Qiagen, Hilden, Germany) and miRNAs in EVs was normalized to their secreted cells (UCMSCs) and represented by the ΔCt values, with a higher ΔCt value representing a less selective sorting of this miRNA into EVs and vice versa. For chondrocyte mRNAs analysis, cells were cultured for one week under treatment as described in Table 1 , and chondrocyte RNAs were isolated as above. cDNA was prepared using SuperScript™ IV Reverse Transcriptase (Thermo Scientific, Massachusetts, USA), and step by step was performed according to the manufacturer’s protocol. cDNA products were then subjected to qPCR reaction, using specific-designed primers that targeted chondrocyte RNAs, including normal chondrocyte markers of Collagen type II (COL2A1), Cartilage oligomeric matrix protein (COMP), and Aggrecan (ACAN) and hypertrophic chondrocyte markers of Collagen type I (COL1A1) and Runt-related transcription factor 2 (RUNX2), and GAPDH as an internal control (primer sequences were listed in Supplementary Table 1 ). 2-ΔΔCt method was applied to calculate the relative fold gene expression of samples.
Human articular chondrocytes were seeded (2,500 cells in each well of 96-well-plate) and incubated in media as listed in Table 1 . No-EV was used as the control. Cells were incubated at 37°C and 5% CO2 for 48 hours to proliferate. The cell proliferation rate was assessed by performing a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay (Abcam, Cambridge, UK) following the manufacturer’s protocols. The proliferation rate was equivalent to the relative absorbance measured at 562 nm (SpectraMax M3, Molecular Devices, California, USA) at time points of 0 hours (as used for normalization) and 48 hours. The proliferation rate was calculated based on the OD values obtained from two time points.
Human articular chondrocytes were cultured in a 24-well plate with a density of 1.05 × 105 cells/well at 37°C and 5% CO2 for attachment. After reaching 100% confluency, cells were then incubated with Mitomycin C (10 μg/mL) for 2 hours to inhibit cell proliferation. A physical scratch was created on the cell attachment layer, and detached cells were removed by washing with media. Treatments were established similarly to proliferation assay ( Table 1 ). Cell migration to close the wound area was captured by an inverted microscope at multiple time points. The wound area was measured using ImageJ software (version 1.48) and calculated for the closure percentage over time, which represents the rate of cell migration.
The statistical analysis was performed on GraphPad Prism 9 (GraphPad Software, California, USA) using One-Way and Two-Way ANOVA, and Tukey HSD tests. The statistical significance was defined as a p-value < 0.05. All data were shown as means ± SD of three biological replicates.
The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.
The conception and design of the study: UT, LN, ThN, HD, and XHN. Analysis and interpretation of data: UT, LN, ThN, HD, XHN, HD, TT, TN, CD, and H-XD. Manuscript drafting: ThN, HD and CD. Manuscript revising: UT, XHN. Final approval: UT. All authors contributed to the article and approved the submitted version.
This project was funded by Vietnam Ministry of Health with decision number 2575/QĐ-BYT (20/06/2019) and Vinmec Joint Stock Company (ISC.18.07). This source has no involvement in the study design, collection, analysis and interpretation of data in the writing of the manuscript and in the decision to submit the manuscript of publication.
We thank the doctors and nurses from Hanoi Medical Hospital for cartilage collection.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647025 | Mona Khan,Marnick Clijsters,Sumin Choi,Wout Backaert,Michiel Claerhout,Floor Couvreur,Laure Van Breda,Florence Bourgeois,Kato Speleman,Sam Klein,Johan Van Laethem,Gill Verstappen,Ayse Sumeyra Dereli,Seung-Jun Yoo,Hai Zhou,Thuc Nguyen Dan Do,Dirk Jochmans,Lies Laenen,Yves Debaveye,Paul De Munter,Jan Gunst,Mark Jorissen,Katrien Lagrou,Philippe Meersseman,Johan Neyts,Dietmar Rudolf Thal,Vedat Topsakal,Christophe Vandenbriele,Joost Wauters,Peter Mombaerts,Laura Van Gerven | Anatomical barriers against SARS-CoV-2 neuroinvasion at vulnerable interfaces visualized in deceased COVID-19 patients | 10-11-2022 | COVID-19,SARS-CoV-2,Coronavirus,Delta,Omicron,olfactory,olfactory sensory neuron,perineurial olfactory nerve fibroblast,olfactory bulb,leptomeninges,brain parenchyma,frontal lobe,neurotropism,neuroinvasion,blood-brain barrier,Virchow-Robin space,glia limitans perivascularis,RNAscope | Can SARS-CoV-2 hitchhike on the olfactory projection and take a direct and short route from the nose into the brain? We reasoned that the neurotropic or neuroinvasive capacity of the virus, if it exists, should be most easily detectable in individuals who died in an acute phase of the infection. Here, we applied a postmortem bedside surgical procedure for the rapid procurement of tissue, blood, and cerebrospinal fluid samples from deceased COVID-19 patients infected with the Delta, Omicron BA.1, or Omicron BA.2 variants. Confocal imaging of sections stained with fluorescence RNAscope and immunohistochemistry afforded the light-microscopic visualization of extracellular SARS-CoV-2 virions in tissues. We failed to find evidence for viral invasion of the parenchyma of the olfactory bulb and the frontal lobe of the brain. Instead, we identified anatomical barriers at vulnerable interfaces, exemplified by perineurial olfactory nerve fibroblasts enwrapping olfactory axon fascicles in the lamina propria of the olfactory mucosa. | Anatomical barriers against SARS-CoV-2 neuroinvasion at vulnerable interfaces visualized in deceased COVID-19 patients
Can SARS-CoV-2 hitchhike on the olfactory projection and take a direct and short route from the nose into the brain? We reasoned that the neurotropic or neuroinvasive capacity of the virus, if it exists, should be most easily detectable in individuals who died in an acute phase of the infection. Here, we applied a postmortem bedside surgical procedure for the rapid procurement of tissue, blood, and cerebrospinal fluid samples from deceased COVID-19 patients infected with the Delta, Omicron BA.1, or Omicron BA.2 variants. Confocal imaging of sections stained with fluorescence RNAscope and immunohistochemistry afforded the light-microscopic visualization of extracellular SARS-CoV-2 virions in tissues. We failed to find evidence for viral invasion of the parenchyma of the olfactory bulb and the frontal lobe of the brain. Instead, we identified anatomical barriers at vulnerable interfaces, exemplified by perineurial olfactory nerve fibroblasts enwrapping olfactory axon fascicles in the lamina propria of the olfactory mucosa.
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the infectious agent responsible for the COVID-19 pandemic. Acute respiratory failure is often the cause of death. But, despite the name of the viral agent, COVID-19 is not uniquely a respiratory disease. , Of relevance here, a plethora of neurological symptoms has become recognized that cannot be linked to the acute respiratory syndrome or be explained by iatrogenic causes. Much attention has been given to the hypothesis that SARS-CoV-2 infects neurons or other cell types in the nervous system (neurotropism) or invades the nervous system (neuroinvasion) and thereby causes pathology. , , Prominent among the neurological manifestations is olfactory dysfunction. Anosmia, the loss of smell, is a common symptom of COVID-19, particularly in the early phases of the pandemic but less so with the Omicron variant. , , There are only a few millimeters between the olfactory mucosa in the upper part of the nasal cavity and the olfactory bulb at the base and front of the cranial cavity. From the get-go, the olfactory projection has been a prime suspect for offering the virus a direct and short route from the nose into the brain. In one scenario, SARS-CoV-2 would infect olfactory sensory neurons (OSNs) at the level of the olfactory epithelium; virions would make it to the olfactory bulb through or along olfactory axon fascicles coursing within the lamina propria of the olfactory mucosa; the virus would spread from the olfactory bulb to the rest of the brain. This hypothetical scenario would explain both the olfactory dysfunction and other neurological manifestations. A major technical hindrance for histological studies of the olfactory system in humans, living or deceased, is the rapid procurement of tissue samples of suitable quality and unambiguous identity. The olfactory mucosa is an archipelago of islands within the respiratory mucosa. It is not possible to harvest samples of pure olfactory mucosa, let alone pure olfactory epithelium. Furthermore, olfactory bulb biopsies cannot be taken from living patients due to the intracranial position and devastating consequences of the intervention. Adapting an endoscopic technique of skull base surgery, we developed a postmortem bedside surgical procedure to rapidly harvest tissue samples of respiratory mucosa, olfactory cleft mucosa, and frontal brain lobe, as well as whole olfactory bulbs. We reported on a first cohort of 85 cases, comprised of 70 COVID-19 patients infected with non-variants of concern or the Alpha variant and 15 non-infected control patients. By combining the RNAscope platform of ultrasensitive single-molecule fluorescence RNA in situ hybridization with fluorescence immunohistochemistry (IHC), we identified sustentacular cells as the major target cell type for SARS-CoV-2 in the olfactory mucosa. We failed to find evidence for infection of OSNs. Here, we included a second cohort of 53 cases, comprised of 45 COVID-19 patients who died a few days after diagnosis of infection with the Delta, Omicron BA.1, or Omicron BA.2 variants and 8 non-infected control patients. Arguably, SARS-CoV-2 would have the greatest chance of displaying its hypothetical neurotropic or neuroinvasive capacity in highly vulnerable patients with a colossal failure of the host defense and the gravest outcome of all. But, in this second cohort as well as in the first cohort, we failed to find evidence for neurotropism and neuroinvasion along the olfactory projection and in the frontal lobe of the brain. We discovered that a poorly characterized cell type, the perineurial olfactory nerve fibroblasts (pONFs), forms a hitherto unrecognized anatomical barrier against SARS-CoV-2 virions.
The study concept of ANOSMIC-19 (ANnalyzing Olfactory dySfunction Mechanisms In COVID-19) is based on a 24/7 workflow that is initiated by a health care worker on an intensive care unit or a ward placing a call to a team of Ear, Nose, and Throat surgeons shortly after death of a COVID-19 patient. A team is dispatched to the bed and brings along a mobile unit consisting of a monitor, light source, camera, and endoscopic equipment. The surgeons then promptly perform the bedside surgical procedure. Retrospectively, we refer to the cohort described in Khan et al. as “cohort-I.” In the new cohort, which we refer to as “cohort-II,” we extended the workflow to include, in most cases, three additional types of samples: a sample of cerebrospinal fluid (CSF) extracted from the cisterna magna, a blood sample drawn from a femoral vein, and several nasopharyngeal swabs taken under endoscopic guidance (Figure 1A). We preferentially included patients who died within ∼14 days after diagnosis of COVID-19 by quantitative reverse-transcription polymerase chain reaction (qRT-PCR), hereafter abbreviated as “days after diagnosis.” Cohort-II is comprised of 45 patients who died from COVID-19 (30 or 67%) or with COVID-19 (15 or 33%) in major hospitals in Leuven (n = 38), Brussels (n = 4), and Bruges (n = 3) between August 2021 and May 2022, and 8 non-infected control patients who died of unrelated causes in Leuven between September 2021 and March 2022 (Figures 1B, S1, and S2). The median of the postmortem interval (PMI) of the COVID cases was 109 min. Cohort-II cases are numbered COVID #71 to #115 and CONTROL #17 to #24. We determined that COVID #71 through #94 were infected with the Delta variant and COVID #95 through #115 with the Omicron BA.1 or Omicron BA.2 variants.
The postmortem nasopharyngeal swabs, serum samples, and CSF samples were analyzed with the TaqPath qRT-PCR assay on the same Quantstudio 7 Flex platform, allowing for direct comparison of the cycle threshold (Ct) values as indications of the viral load. All 38 nasopharyngeal swabs of the Leuven cases tested positive, with Ct values ranging from 7.9 to 31.0 (Figure 2A). Of the 36 available serum samples of the Leuven cases, 18 (50%) tested positive, with Ct values ranging from 22.3 to 35.4 (Figure 2B). Of the 35 available CSF samples of the Leuven cases, 2 (6%) tested positive (Figure 2C); as both cases were RNAemic, we cannot exclude that a traumatic puncture led to contamination with blood. Serum anti-S immunoglobulin G (IgG) antibody titers varied from below the detection limit (<50 arbitrary units [AU]/mL) to above the limit of quantification (>80,000 AU/mL), indicating a broad range of humoral responses against vaccine-derived spike protein and/or virus-derived spike protein (Figures 2D and S3). Seroconversion for anti-N IgG antibodies was observed in six serum samples. Taken together, the high viral loads of the nasopharyngeal swabs and the frequent occurrence of an RNAemic state are consistent with an acute phase of infection.
Cohort-I included patients who died between May 2020 and April 2021, an era in which COVID-19 vaccines were not yet authorized or the Belgian vaccination campaign had just been initiated. By contrast, 37 of the 45 COVID cases (82%) in cohort-II were vaccinated. Four (9%) were not vaccinated, and the vaccination status of 4 patients (9%) was unknown. The vaccines most commonly administered were mRNA vaccines tozinameran/Comirnaty and elasomeran/Spikevax, and less frequently administered were vector vaccines ChAdOx1-S/Vaxzevria and Ad26.COV2-S/Janssen Jcovden. An infection with SARS-CoV-2 after vaccination is called a “breakthrough infection,” a term that encompasses a spectrum of outcomes ranging from asymptomatic infection to death. , A challenging distinction is that some of the patients who die during a breakthrough infection die from COVID-19 whereas others die with COVID-19. Pragmatically, we coin the term “fatal breakthrough infection” (FBI) for vaccinated patients in our cohorts who died from COVID-19. Of the 37 vaccinated cases in cohort-II, 25 (68%) are, to the best of our clinical judgment, FBI cases. In cohort-I, only COVID #68 was vaccinated, and this patient was judged an FBI case.
SARS-CoV-2 is a positive-sense single-stranded RNA virus. Positive sense means that the viral genomic RNA can be translated directly into protein. A single SARS-CoV-2 virion contains a single positive-sense full-length genomic RNA molecule. We performed confocal imaging of sections stained with a combination of fluorescence RNAscope, which visualizes a single RNA molecule as a dot or “punctum” (plural “puncta”), and fluorescence IHC, which visualizes an antigen as an immunoreactive (IR) signal. Our panel of viral RNAscope probes consists of SARS-CoV-2-N (nucleocapsid; giving rise to puncta referred to as N puncta), SARS-CoV-2-S (spike; S puncta), SARS-CoV-2-orf1ab (open reading frames 1a and 1b; orf1ab puncta), SARS-CoV-2-N-sense (N-sense puncta), SARS-CoV-2-S-sense (S-sense puncta), and SARS-CoV-2-orf1ab-sense (orf1ab-sense puncta). The sense probes detect negative-sense full-length genomic RNAs and negative-sense subgenomic RNAs, which are produced as intermediates during the viral life cycle but are not incorporated into virions released from cells. , RNAscope puncta for sense probes therefore reflect ongoing viral replication, and sense puncta reside in a characteristic perinuclear position. The presence of extracellular N puncta, S puncta, or orf1ab puncta combined with the absence of sense puncta reflects the presence of extracellular virions released from cells. Nucleocapsid-IR signal diffusely fills an infected cell and outlines its contours, thereby facilitating cell type identification.
Ciliated cells are the main target cell type in the respiratory epithelium for non-variants of concern and the Alpha variant. Ciliated cells can be identified by puncta for FOXJ1, which encodes a transcription factor involved in ciliogenesis, and by KRT8-IR signal. We detected viral RNA puncta and nucleocapsid-IR signal in respiratory mucosa samples of 41 of the 45 COVID cases of cohort-II (91%). We refer to such cases as “informative” and to the other cases as “non-informative.” In the respiratory epithelium of COVID #87, a Delta FBI patient who died 3.7 days after diagnosis, many KRT8-IR cells harboring FOXJ1 puncta also harbor N puncta (Figure 3A); nearly all nucleocapsid-IR cells harbor N-sense puncta and orf1ab-sense puncta, reflecting ongoing viral replication (Figure 3B). In COVID #95, an Omicron BA.1 FBI patient who died 1.7 days after diagnosis, cells harboring FOXJ1 puncta harbor N-sense puncta and are situated along a stretch of the respiratory epithelium that is covered by a thick layer of secreted mucus containing IR signal for MUC5AC, a gel-forming glycoprotein secreted by goblet cells (Figure 3C). In COVID #107, an Omicron BA.2 patient who died 1.6 days after diagnosis, many cells with the typical position and shape of ciliated cells are nucleocapsid-IR, and all of these harbor S-sense puncta perinuclearly (Figure 3D). In COVID #101, an Omicron BA.2 patient who died 4.5 days after diagnosis, densely packed S puncta occur in nucleocapsid-IR cells, with IR signal diffusely filling the infected cells and outlining their contours including their cilia (Figure 3E). In this case, some KRT8-IR cells at the apical edge of a crypt or gland duct harbor N puncta and a blob of secreted mucus is MUC5AC-IR (Figure 3F). Deeper in this crypt or gland duct, there are no intracellular N puncta and the surface of the epithelium is covered with clusters of extracellular N puncta (Figure 3G); similar clusters of extracellular S puncta are nucleocapsid-IR (Figure 3H). A massive and diffuse infection across a broad swath of the respiratory epithelium of COVID #101 is revealed in a tiled confocal image: nearly every nucleocapsid-IR cell harbors N-sense puncta and orf1ab-sense puncta (Figure 3I). Likewise, Omicron BA.1 swept across a broad swath of the respiratory epithelium of COVID #104, who died 1.6 days after diagnosis (Figure 3J). Taken together, Delta, Omicron BA.1, and Omicron BA.2 can mount a massive attack on ciliated cells in the respiratory epithelium. The extracellular N puncta, S puncta, and nucleocapsid-IR signal that we captured in COVID #101 reflect extracellular virions released into the external milieu.
We found in a subset of cases that SARS-CoV-2 invaded the interstitium of the lamina propria of the olfactory mucosa. Interstitial invasion raises the question of whether virions can infect cells of the olfactory projection at the level of the lamina propria, or hitchhike onto the olfactory projection, and then invade the brain. Before presenting these cases, we describe the anatomical relationships of OSN axons with other cellular components after they exit the olfactory epithelium and enter the lamina propria. Particularly didactic in this regard is COVID #105, in whom no viral presence was detected. Figure 4A shows a continuous swath of olfactory epithelium rich in cells harboring puncta for OMP, which encodes the olfactory marker protein, a classical marker for mature OSNs. Each mature OSN projects a single axon through a perforation of the basement membrane into the superficial lamina propria, where it is received by olfactory ensheathing cells (OECs), a type of glia cells unique to the olfactory projection. , , OECs can be identified with S100B-IR or GFAP-IR signal. In the superficial lamina propria, the OECs run criss-cross and are cut in various planes. Progressively, OSNs and OECs together form olfactory axon fascicles of increasing diameter and become enwrapped by pONFs. , , , , , , This cell type can be identified in human tissue sections with p75-IR signal, with p75 as abbreviation for p75NTR, the p75 neurotrophin receptor also known as low-affinity nerve growth factor receptor. , Figure 4B shows two major cell types of the olfactory epithelium: sustentacular cells, harboring UGT2A1 puncta, and OSNs, labeled with TUBB3-IR signal , ; in the superficial lamina propria, OSN axons are intimately associated with OECs. Deeper in the lamina propria, transversely cut olfactory axon fascicles are enwrapped individually by pONFs (Figure 4C). The difference in pONF enwrapment of olfactory axon fascicles in superficial vs. deep lamina propria is prominent in Figure 4D. Closer to the cribriform plate, olfactory axon fascicles are thicker, and all have become enwrapped by pONFs (Figure 4E). The OSN provenance of the enwrapped cellular structures can be confirmed with OMP-IR signal (Figures 4F and 4G). Taken together, the fuzzy and loose configuration of the progressive fasciculation of OSN axons, ensheathment by OECs, and enwrapment by pONFs poses a vulnerability to the olfactory pathway and the brain, which is only a few millimeters away.
We discovered a pattern of interstitial SARS-CoV-2 spreading by virtue of the RNAscope-based visualization of extracellular virions, consistent with the well-documented hallmark of RNAscope to visualize single RNA molecules as puncta in light microscopy. Figure 5 shows that pONFs form an anatomical barrier against such virions. In reference to perineurial ONFs, we refer to seven cases as “perineurial cases": COVID #68 (Alpha, FBI) of cohort-I, and #87 (Delta, FBI), #89 (Delta, FBI), #90 (Delta), #94 (Delta, FBI), #108 (Omicron BA.1, unvaccinated), and #110 (Omicron BA.1) of cohort-II. In COVID #68, who died 5 days after diagnosis, a collection of N puncta in the interstitium does not intermingle with TUBB3-IR OSN axons (Figure 5A); trains of N puncta remain outside of a fascicle of TUBB3-IR OSN axons tightly associated with GFAP-IR OECs (Figure 5B); and a swarm of N puncta stays clear of the enwrapment by p75-IR pONFs (Figure 5C). In COVID #87, who died 3.7 days after diagnosis, a thick olfactory axon fascicle was cut longitudinally over a distance of ∼1.2 mm and is enwrapped without interruption by a thin layer of p75-IR pONFs; it is accompanied by an array of N puncta in the interstitium (Figure 5D), but the fascicle is devoid of N puncta (Figure 5E). In COVID #89, who died 5.1 days after diagnosis, N puncta are dispersed across the interstitium of the lamina propria but not within olfactory axon fascicles identified with S100B-IR signal of OECs and p75-IR signal of pONFs (Figures 5F and 5G) or with OMP-IR signal and TUBB3-IR signal of OSN axons (Figure 5H). In COVID #94, who died 8.3 days after diagnosis, dense accumulations of N puncta are mutually exclusive with TUBB3-IR OSN axons (Figure 5I). In COVID #108, who died 3.9 days after diagnosis, an ant-like column of N puncta escorts a thick fascicle of TUBB3-IR axons without invading it (Figure 5J). We confirmed and extended these findings with RNAscope for orf1ab (Figures S4A, S4B, S4D–S4G, and S4I) and S (Figures S4C, S4F, and S4H–S4J): extracellular puncta do not invade TUBB3-IR axon fascicles and stay clear of the enwrapment by p75-IR pONFs. A haze of extracellular nucleocapsid-IR signal colocalizes with viral puncta (Figures S4E, S4F, and S4I). There are no N-sense puncta (Figures S4A, S4B, and S4J) or S-sense puncta (Figures S4D and S4G). Taken together, pONFs form an anatomical barrier against viral invasion of the olfactory projection at a vulnerable interface in the lamina propria of the olfactory mucosa.
The term “fila olfactoria” is used for olfactory axon fascicles penetrating through foramina of the cribriform plate or coursing within the cranial cavity. These thread-like structures are macroscopically visible (see Video S1 at 1′40″ of Khan et al.). The external surfaces of the fila olfactoria are contiguous with the leptomeninges (pia and arachnoid) that snugly cover the olfactory bulb. This anatomical contiguity raises the question of whether virions could sneak up from the interstitium of the lamina propria to the leptomeninges of the olfactory bulb by hitchhiking along the olfactory projection and then invade the parenchyma of the olfactory bulb. We reported the presence of viral RNA in the leptomeninges of the olfactory bulb in 11 of the 30 (37%) informative cases of cohort-I. We now report a leptomeningeal phenotype in 12 of the 41 informative cases (29%) of cohort-II. Suspiciously, six of the seven perineurial cases are also leptomeningeal. In COVID #89, N puncta are dispersed across a SSTR2A-IR swath of leptomeninges but do not occur in the parenchyma (Figure 6A). In COVID #108, densely packed N puncta occur in pia mater abutting virus-free parenchyma, in which TUBB3-IR OSN axons terminate in globose structures, the glomeruli (Figures 6B and 6C). Abundant N puncta in the leptomeninges are associated with blood vessels identified with PECAM1 puncta in endothelial cells (Figure 6D). Despite the close proximity to SSTR2A-IR leptomeningeal tissue, N puncta do not occur within TUBB3-IR fila olfactoria or glomeruli (Figure 6E). Sprinkles of N puncta accompany a TUBB3-IR filum olfactorium and are spread across the abutting leptomeninges (Figure 6F). Taken together, the parenchyma of the olfactory bulb is spared from viral invasion.
We harvested brain tissue close to the olfactory bulb, typically the gyrus rectus and sometimes the gyri orbitales of the frontal lobe, in 112 of the 115 cases of cohort-I and cohort-II. We identified the sporadic occurrence of extracellular N or S puncta in 5 of the 112 cases: COVID #60 (non-variant of concern, unvaccinated), #87 (Delta, FBI), #89 (Delta, FBI), #108 (Omicron BA.1, unvaccinated), and #110 (Omicron BA.1). All five cases were RNAemic perimortem. The puncta occurred intravascularly or perivascularly, but not in the parenchyma. The blood-brain barrier includes the layer of endothelial cells of cerebral blood vessels connected with tight junctions (Figure 7A). The Virchow-Robin space (VRS) is a CSF-filled perivascular space that exists along the larger cerebral blood vessels but not around smaller blood vessels such as capillaries. , The glia limitans perivascularis covers the entire cerebral vasculature: all arterioles, capillaries, and venules within the brain parenchyma are surrounded by vascular endfeet of astrocytes. The tight junctions among endothelial cells can be visualized as short TJP1-IR stripes, for tight junction protein-1 (Figure 7B). Endothelial cells harbor PECAM1 puncta, and astrocytes, including their endfeet, can be identified with AQP4-IR signal, for aquaporin-4 (Figure 7C). COVID #60 had a malignant brain tumor and died 40 h after diagnosis due to increased intracranial pressure that was treated with high doses of corticosteroids; a serum sample taken a few hours prior to the time of death had a Ct of 33.6. A few N puncta are visible within the lumen of a blood vessel outlined with CD31-IR endothelial cells harboring PECAM1 puncta (Figure 7D), and a compact cluster of N puncta appears to have leaked through the endothelial wall (Figure 7E). In COVID #87, N puncta occur intravascularly and perivascularly but remain confined to the VRS and do not reach beyond the AQP4-IR ring of astrocytic endfeet (Figures 7F and 7G). In COVID #89 (Figure 7H) and in COVID #110 (Figure 7I), N puncta occur within the lumen of a blood vessel but not beyond the AQP4-IR ring. In COVID #108, puffs of N or S puncta appear to have leaked through the endothelial wall identified with PECAM1 puncta (Figures 7J–7L), TJP1-IR signal (Figure 7J), and CD31-IR signal (Figure 7K) but did not proceed into the parenchyma (Figure 7L). Taken together, we failed to find evidence for viral invasion of the parenchyma of the frontal lobe in 112 of 112 (100%) cases.
Finally, we asked whether SARS-CoV-2 virus could be grown in culture from CSF samples extracted postmortem from the cisterna magna (Figure 8A). As substrate we used inserts of human airway epithelial cell cultures with an air-liquid interface. These differentiated primary cultures possess the architecture and cellular complexity of the epithelium of the human respiratory tract and serve as assays for profiling antiviral drugs against SARS-CoV-2, including Delta and Omicron BA.1. An inoculum is applied to the apical side of an insert; the apical side is washed with medium at several time points; RNA is extracted from the wash fluids; and SARS-CoV-2 RNA is quantified by qRT-PCR. In Figure 8B, we verified that CSF does not inhibit growth of SARS-CoV-2 in this platform by adding CSF from CONTROL #19 to an inoculum of Delta strain hCoV-19/Belgium/rega-7214/2021. In parallel, we tested CSF samples of nine Delta cases, including COVID #83, whose CSF sample had a Ct of 31.1; no virus replication was detected. In a second experiment (Figure 8C), virus could be grown from the nasopharyngeal swab, but not from the CSF sample, of COVID #99, and neither could it be grown from CSF samples of nine other Delta and Omicron BA.1 cases, including COVID #87 and #89, whose frontal lobe samples contained N puncta. In a third experiment (Figure 8D), virus could be isolated from nasopharyngeal swabs of four cases, except for COVID #103, who died 14.1 days after diagnosis, but not from paired CSF samples, including COVID #108, whose frontal lobe sample contained N puncta. Finally, virus grew from a freshly taken nasopharyngeal swab, but not from the paired CSF sample of COVID #96 (Figure 8E). Taken together, in none of 25 COVID cases could we grow SARS-CoV-2 from CSF samples, including from three cases with viral puncta in the frontal lobe sample. By contrast, paired nasopharyngeal swabs routinely gave rise to virus production. In summary, based on the results of this study and Khan et al., we conclude that SARS-CoV-2 does not infect OSNs, olfactory bulb neurons, or neurons in the frontal lobe; it does not invade olfactory axon fascicles, fila olfactoria, the parenchyma of the olfactory bulb and of the frontal lobe; and virions are not recoverable from CSF samples.
In this virocentric view of COVID-19, we visualized infected cells, ongoing viral replication, and extracellular virions in postmortem tissue samples of patients who died during an acute phase of SARS-CoV-2 infection.
Cohort-II of ANOSMIC-19 comprises 45 patients who died from or with COVID-19 shortly after diagnosis in three major hospitals in Belgium over a period of eight months in 2021 and 2022 that spanned the Delta, Omicron BA.1, and Omicron BA.2 waves of the pandemic in Belgium. Our method of procuring postmortem tissue samples takes place at the bedside soon after the death of the patient, with a rapid endoscopic method that allows for precise harvesting of endonasal and intracranial tissue samples. Consistent with the short PMIs, none of the 45 COVID cases and 8 control cases had to be excluded because of poor tissue quality. We report the period until death that elapsed from the time the positive nasopharyngeal swab was taken rather than from the onset of symptoms, which is an unreliable measure, particularly in patients who were already in ill health prior to the infection.
Our qRT-PCR assays revealed RNAemia in 50% of the postmortem serum samples. RNAemia, the presence of viral RNA in the blood, may reflect viremia, the presence of virus in the blood. , This frequent observation of RNAemia and the close association of N and S puncta with blood vessels in the frontal lobe samples of our five COVID cases suggest RNAemia as an explanation for positive qRT-PCR results on homogenized brain samples in some of the literature on autopsies. Blood is present in all tissues, and to various extents. Unless a blood sample from around the time of death is available, caution must be exercised when interpreting claims of neuroinvasion that are largely based on such qRT-PCR analyses. An early claim of neuroinvasion proposed the olfactory pathway as a port of central nervous system entry. But, the sole antibody for a viral antigen that was used in this study, 3A2, gives similar IR signals in control cases. , , , Throughout cohort-I and cohort-II, we have included control cases in parallel, following the same protocols of tissue sample procurement, fixation, processing, staining, and imaging. This two-armed study design safeguards us against false-positive signals during the course of a multi-year project. The fluorescence RNAscope platform has a high signal-to-noise ratio, yielding a high specificity. The sense probes reflect ongoing viral replication, and sense puncta have a characteristic perinuclear location. The combination of nucleocapsid being the most abundant SARS-CoV-2 protein, being a cytosolic protein as opposed to membrane-bound spike, and diffusely filling infected cells and outlining their contours, underlies the strong and distinctive nucleocapsid-IR signals in our confocal images. With the dual visualization of RNAscope puncta and nucleocapsid-IR signal, a given cell type or a particular cell can be unmistakably identified as a target cell type or infected, respectively. Applying these stringent criteria, we failed to find evidence for neurotropism—defined as the ability of a virus to infect and replicate in cells of the nervous system—in OSNs, olfactory axon fascicles, fila olfactoria, and the parenchyma of the olfactory bulb and of the frontal lobe. What about neuroinvasiveness, defined as the ability of a virus to enter the nervous system? Could virions invade the olfactory projection and the brain without infecting cells? In other words, could SARS-CoV-2 be neuroinvasive but not neurotropic? The pristine condition of our tissue samples, the power of the RNAscope platform, and the precision of confocal imaging enabled us to light-microscopically visualize extracellular virions released from infected cells and spreading interstitially through tissues. The smallest extracellular puncta may well represent single virions, consistent with the extensively documented hallmark of RNAscope visualizing single RNA molecules as puncta. By using a battery of RNAscope probes and an anti-nucleocapsid antibody, we failed to find evidence for neuroinvasion of the parenchyma of the olfactory bulb and of the frontal lobe. But the human brain is a large and anatomically highly structured organ, and we reasoned that we should also look for the presence of infectious virions in the CSF. The brain of an adult human is bathed in ∼150 mL of CSF, of which we tapped ∼5 mL from the cisterna magna, which communicates with the other ventricles. We failed to find evidence for infectious virions in CSF samples, including from one COVID case with a positive qRT-PCR result of the CSF sample (#83) and three COVID cases with N puncta in frontal lobe samples (#87, #89, and #108). By contrast, SARS-CoV-2 could be routinely grown from paired nasopharyngeal swabs. Taken together, we failed to find evidence for neurotropism and neuroinvasion in two successive cohorts comprised of a total of 115 patients infected with non-variants of concern, Alpha, Delta, Omicron BA.1, or Omicron BA.2 and surviving between 0 and 40 days after diagnosis. The colossal failure of host defense would afford the most permissive conditions for viral replication and spreading in these individuals. We submit that our analyses of these cases represent the most stringent test imaginable to date for the neurotropic and neuroinvasive capacity of SARS-CoV-2. Admittedly, the absence of evidence does not equal the evidence of absence. It is ultimately not possible to prove a negative in science and in medicine. Certain types of neurons may eventually be found to be infected, perhaps in a subset of patients, in certain disease courses or phases, or elsewhere in the brain. A future variant may inflict (collateral) damage to the brain by newly acquired neurotropic or neuroinvasive properties without suffering disadvantage in the incessant evolutionary race to ever-greater transmissibility, fitness, and immune evasion.
By analogy with peripheral nerves, which have an endoneurium, a perineurium, and an epineurium, pONFs are said to be in a perineurial position with regard to the olfactory nerve. These cells enwrap, as one or a few layers, olfactory axon fascicles consisting of OSN axons ensheathed by OECs. The term “fibroblast” may well reflect the depth of our lack of knowledge about this cell type rather than a cell-biological characteristic. Ultrastructurally, rat pONFs are extremely thin elongated cells with nuclei that reach 15 μm in length. In the seven perineurial cases of cohort-I and cohort-II, the lamina propria of the olfactory mucosa contains extracellular N, S, and orf1ab puncta, reflecting virions spreading within the interstitium. Viral invasion of the lamina propria may occur when the olfactory epithelium is structurally damaged due to massive infection of sustentacular cells, ensuing desquamation, and erosion of the basement membrane. Alternatively, and not mutually exclusively, virions may invade the lamina propria via the hematogenous route. Indeed, the five perineurial cases for whom blood samples were available were RNAemic. Collections of extracellular N puncta assume several geometric configurations. They may follow the paths of lowest resistance and be propelled by bulk flow toward the cranial cavity. They approach and come so close to olfactory axon fascicles that they seem to trail their contours, but they do not infect pONFs and do not penetrate beyond the p75-IR enwrapment. The mechanistic basis of the pONF barrier remains to be determined. We speculate that this barrier may also be effective against some of the many other pathogens that infect the nasal mucosa and could threaten the brain.
We detected extracellular N and S puncta in the leptomeninges (pia mater and arachnoid) that cover the surface of the olfactory bulb. These leptomeninges are anatomically contiguous with the external surfaces of olfactory axon fascicles and fila olfactoria. We reported a leptomeningeal phenotype in 11 of the 30 informative cases of cohort-I in Khan et al. and now add 12 of the 41 informative cases of cohort-II for a total of 23 of the 71 informative cases (32%). The absence of sense puncta in the leptomeninges argues against ongoing viral replication at these sites. Suspiciously, six of the seven perineurial cases are also leptomeningeal; the case with the weakest perineurial phenotype (COVID #90) is not leptomeningeal. It is tempting to speculate that perineurial cases and leptomeningeal cases belong to the same spectrum, with virions hitchhiking along the external surface of the pONF enwrapment into the leptomeninges covering the olfactory bulb. The fate of the leptomeningeal virions is uncertain, but they do not invade the parenchyma of the olfactory bulb. The leptomeninges-parenchyma interface is vulnerable, as the olfactory bulb contains the second-order neurons of the olfactory pathway, the mitral and tufted cells, which project their axons via the olfactory tract to the piriform cortex and many other brain regions. Hence, infection of these neurons would be an effective way for SARS-CoV-2 to spread throughout the brain, but it is a road not taken. Our findings of leptomeningeal virions in 32% of informative cases are consistent with those of a neuropathological study that examined 16 brain regions in COVID-19 decedents by qRT-PCR for several SARS-CoV-2 gene sequences. In 8 out of 21 cases (38%), viral RNA was detected in olfactory bulb samples, and in some of these cases also in amygdala and entorhinal area, brain regions closely connected to the olfactory bulb, but typically not elsewhere in the brain. It would be interesting to determine with RNAscope in cases of the Beach cohort whether viral RNA is detectable in the leptomeninges covering the olfactory bulb vs. in the parenchyma.
In a mere 5 of 112 cases of cohort-I and cohort-II for whom frontal lobe tissue samples were available, we detected viral signal—sporadically, as clusters or puffs of extracellular N puncta or S puncta, the proverbial needles in a haystack. Viral puncta did not present within the parenchyma but were confined to the lumen of a blood vessel or appeared to have leaked through the endothelial layer. Viral puncta did make it as far as the VRS, but not beyond the glia limitans perivascularis, which is formed by AQP4-IR vascular endfeet of astrocytes and covers the cerebral vasculature along its entire length of several hundred kilometers. Strikingly, the antibody response to vaccine-derived spike protein and/or virus-derived spike protein was close to non-existent in these five cases: COVID #60, #89, #108, and #110 had anti-S IgG titers below the limit of detection (<50 AU/mL), and COVID #87 had a very low titer of 80.2 AU/mL. Moreover, there was no seroconversion for anti-N IgG in these five cases. Thus, even in the context of an abysmal failure of the humoral response to the vaccine and/or the virus, SARS-CoV-2 failed to invade the parenchyma of the olfactory bulb and of the frontal lobe. Regular glymphatic cleansing of the brain may efficiently dilute virions that made it to the CSF or entered the brain parenchyma. Meningeal lymphatic vessels may help dispose of SARS-CoV-2 virions. On the other hand, the integrity of the anatomical barriers may be compromised and breaches resulting in neuroinvasion may be possible. The blood-brain barrier is leaky in neurodegenerative disorders such as Alzheimer’s disease. Interestingly, perineurial case #94, whose CSF sample tested positive by qRT-PCR, had Alzheimer’s disease. Microbleeds or fleeting events of microneuroinvasion may go unnoticed in histological studies like ours but might be sufficient to trigger a sequence of pathobiological events culminating in chronic neuroinflammation and lasting dysfunction such as some of the neurological manifestations of long COVID.
Similar to sustentacular cells in the olfactory epithelium playing the role of unsung heroes of the sense of smell, perineurial olfactory nerve fibroblasts are another band of unsung heroes: they form a protective anatomical barrier at a vulnerable interface of the olfactory projection. Through mechanisms that remain to be elucidated, this enigmatic cell type appears to seal olfactory axon fascicles hermetically from invasion by SARS-CoV-2 virions only 100 nm in diameter.
The scope of the study was limited to light-microscopically visualizing whether SARS-CoV-2 is capable of neurotropism and neuroinvasion in samples of olfactory cleft mucosa, olfactory bulb, and frontal lobe. We took the viewpoint of the virus and not of the host response for a host of reasons, including the heterogeneity of patients in terms of comorbidities, medical treatments, previous exposures to SARS-CoV-2 (diagnosed or undiagnosed), vaccination regimens, and antibody titers to spike and nucleocapsid antigens. An investigation of neuroinflammation was beyond the scope of the study. Most COVID-19 patients, vaccinated or unvaccinated, recover from their infection, raising the fundamental question as to whether postmortem studies adequately reflect the situation in individuals with a non-fatal disease outcome. This issue will be almost impossible to resolve, as there are no ethically justifiable indications for brain biopsies from living COVID-19 patients. We sampled only the gyrus rectus and gyri orbitales and may have overlooked neurotropism and neuroinvasion elsewhere in the brain. A study of a postmortem cohort cannot be undertaken longitudinally, by its very nature. Instead, we must make inferences from snapshots taken during many singular courses of infection and then attempt to fill in the dots and make associations. It does appear, though, that SARS-CoV-2 is stopped dead in its tracks by several anatomical barriers at vulnerable interfaces, even in extremely weak individuals with an abysmal level of defense who lost the battle.
Further information and requests should be directed to the lead contact, Peter Mombaerts ([email protected]).
This study did not generate new unique reagents.
• Clinical data about the patients are confidential, subject to compliance with applicable personal data protection laws, and not publicly available. • This paper does not report original code. • Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.
The foundation of the study protocol ANOSMIC-19 (ANalyzing Olfactory dySfunction Mechanisms in COVID-19) is the bedside procurement of postmortem tissue samples. This national multicenter study was approved by the Ethical Committees of the University Hospitals Leuven in Leuven, Belgium (S64042), of the General Hospital Sint-Jan Brugge-Oostende AV in Bruges, Belgium (2736), and of the Universitair Ziekenhuis Brussel in Brussels, Belgium (EC-2021-360), and registered on clinicaltrials.gov (NCT04445597). The Ethikrat – Kommission des Präsidenten of the Max Planck Society did not require a separate ethics review by a medical ethics committee (Applications No: 2020_14, 2020_30, and 2020_31). Patients were >18 years old at the time of inclusion. Written informed consent from next of kin was obtained prior to tissue harvesting in accordance with the recommendations of the local Ethical Committee. COVID-19 patients were diagnosed with a SARS-CoV-2 infection by qRT-PCR from a nasopharyngeal swab and died during their subsequent COVID-19 hospitalization. Control patients had a negative qRT-PCR test from a postmortem nasopharyngeal swab taken a few days prior to their time of death and died of other causes than COVID-19. The electronic health records of each patient were retrospectively reviewed and analyzed to obtain information about demographics, comorbidities, disease course, and hospitalization history. The collection, processing, and disclosure of personal data (such as patient demographic, health, and medical information) are subject to compliance with Regulation (EU) 2016/679, also referred to as the General Data Protection Regulation, and the Belgian Law on the protection of natural persons regarding the processing of personal data. Therefore, combinations of data deemed to be identificatory to specific persons cannot be disclosed. For instance, details of the vaccination record of an individual decedent cannot be reported. The narrowing of the time between diagnosis and death to ∼14 days led to the inclusion of a greater proportion of patients from a ward (41 or 91%) than from an ICU (4 or 9%) compared to cohort-I (respectively 46% and 54%).
Comorbidities were categorized in accordance with international recommendations. Overweight is as a body mass index (BMI) >25 kg/m2, and obesity as a BMI ≥30 kg/m2. Presence of diabetes mellitus type 2 includes previously known and newly diagnosed patients, based on Hb1Ac ≥6.5% or active treatment on admission. Former smokers, defined as having ceased smoking >6 months prior to inclusion, are not considered smokers in Figures 1B and S1. Hypertension is defined as grade 1 hypertension, or treatment with antihypertensive drugs. Chronic kidney disease is defined as the presence of kidney damage or a glomerular filtration rate of <60 mL/min/1.73 m2 for >3 months. Chronic lung disease includes obstructive lung disease (chronic obstructive pulmonary disease, asthma), interstitial lung disease, pulmonary fibrosis, and pulmonary hypertension. Cardiovascular disease comprises heart conditions (such as valvular disease, heart failure, arrhythmias, cardiomyopathies, coronary artery disease), cerebrovascular antecedents, and history of pulmonary embolism. Neurodegenerative disease comprises all forms of dementia and progressive cognitive impairment. Patients were considered immunocompromised if one of the following criteria was met: (1) an active oncological condition, defined as presence of a solid tumor or hematologic malignancy <6 months prior to inclusion; (2) immunosuppressive drugs as maintenance therapy, including corticosteroids and chemotherapy; (3) recipient of a solid organ transplant. The cause of death of COVID-19 patients was classified as one out of two categories. (1) Death from COVID-19: hypoxic respiratory failure secondary to SARS-CoV-2 pneumonia or death from COVID-19 sequelae such as acute respiratory distress syndrome, fungal or bacterial superinfection associated with SARS-CoV-2 pneumonia. (2) Death with COVID-19: cause of death not directly related to COVID-19 such as acute cardiac arrest, cerebrovascular accidents, or deterioration of an oncological condition. Cohort-II patients were included between August 2021 and May 2022. During this period, "vaccinated" was defined as having received, at least two weeks earlier, either two doses of tozinameran/Comirnaty (Pfizer-BioNTech), two doses of elasomeran/Spikevax (Moderna), two doses of ChAdOx1-S/Vaxzevria (AstraZeneca), or a single dose of Ad26.COV2-S/Janssen Jcovden (Johnson & Johnson). Several patients received a booster in the fall of 2021 and some a second booster. The term "vaccinated" in this study encompasses all these vaccination histories. Cohort-I cases died between May 2020 and April 2021, at a time when COVID-19 vaccines did not exist, or the Belgian vaccination campaign had just been rolled out. Only one case of cohort-I, COVID #68, was vaccinated. The definition of "vaccinated" will evolve in the future and the definition of "breakthrough infection" will have to be adapted accordingly. Breakthrough infections occur frequently, and some of the patients who die during the course of a breakthrough infection succumb to the SARS-CoV-2 infection as opposed to passing away with it. The term "fatal breakthrough infection" (FBI) that we here coin is pragmatic and concise, and refers to patients who, to the best of the clinical judgment of the physicians who treated or reviewed the patient files, died from COVID-19 instead of with COVID-19 despite being vaccinated. The term FBI does not imply any underlying cause such as the absence of anti-S IgG in serum on admission or postmortem. In the foreseeable future, unvaccinated patients will no longer represent typical COVID-19 patients.
The first step of the extended postmortem bedside surgical procedure is to extract a CSF sample from the cerebellomedullary cistern, also known as the cisterna magna. The cisterna magna is located between the medulla oblongata and the cerebellum in the posterior fossa. To achieve optimal exposure for a cisternal puncture, the decedent was positioned in a supine, anti-Trendelenburg position or turned sideways, depending on the physiognomy of the body. The neck of the patient was positioned in flexion, the occiput was palpated downwards until encountering the spine of the axis vertebra and a spinal needle (22 Gauge, 90 mm) was inserted on the midline in the depression between these two points. The needle was directed upwards toward the midpoint of an imaginary line joining the left and right external auditory meatus, until the needle was felt to pierce the atlanto-occipital ligament, thereby entering the cisterna magna. The needle hub was then inspected meticulously to ascertain that the liquid is clear and colorless, before continuing to sample a syringe of 5 mL CSF. In case of a traumatic puncture (revealed as the presence of blood in the needle hub), the procedure was repeated from the beginning. CSF samples were sent to the National Reference Center for Respiratory Pathogens at the Department of Laboratory Medicine of the University Hospitals Leuven (Leuven, Belgium) for PCR analysis and to the BSL-3 facility of the KU Leuven Rega Institute (Leuven, Belgium) for viral culture.
The second step of the extended postmortem bedside surgical procedure is to draw a blood sample via a femoral puncture or, when available, via an existing central venous line. For the femoral puncture, an 18 Gauge needle was inserted ∼2 cm below the midpoint between the anterior superior iliac spine and the pubic symphysis. Whole blood was collected in 5 mL Serum Separator Tubes (BD Vacutainer). After centrifugation, the serum was stored in sterile tubes and serum samples were sent to the National Reference Center for Respiratory Pathogens at the Department of Laboratory Medicine of the University Hospitals Leuven, (Leuven, Belgium) for qRT-PCR analysis.
The third step of the extended postmortem bedside surgical procedure entails the initiation of the endoscopic endonasal procedure, using a 4 mm 0° endoscope connected with a camera and monitor and a light source (Karl Storz). Prior to cutting and tissue sample harvesting, three nasopharyngeal swabs were taken under endoscopic guidance. A first nasopharyngeal swab was used immediately for a bedside rapid antigen test, using the Panbio Abbott COVID-19 Rapid Test Device (Abbott, REF#41FK10) against SARS-CoV-2 nucleocapsid antigen. A second nasopharyngeal swab was for TaqPath COVID-19 qRT-PCR and whole-genome sequencing at the Belgian National Reference Center for Respiratory Pathogens at the Department of Laboratory Medicine of the University Hospitals Leuven (Leuven, Belgium). A third nasopharyngeal swab was sent to the BSL-3 facility of the KU Leuven Rega Institute (Leuven, Belgium) for viral culture.
The fourth and final step of the extended postmortem bedside surgical procedure consists of two parts: an intranasal and an intracranial part. Intranasal tissue samples of the respiratory mucosa and olfactory cleft mucosa were collected via a classical endoscopic endonasal procedure with cold instruments. The procedure was described in detail in Khan et al. Briefly, to harvest respiratory mucosa, we removed the inferior turbinate and the middle turbinate after cutting their attachments to the lateral nasal wall and nasal roof with Heymann and endoscopic scissors. To harvest olfactory cleft mucosa, an elliptical incision with a sickle or beaver knife was made running over the superior part of the septum, the cribriform plate, and the area of the vertical attachment of the superior turbinate, thus covering the full extent of the olfactory cleft region. A subperiosteal dissection was initiated on the medial side (superior part of the septum) and lateral side (vertical attachment of the turbinates), progressively extending to the center (cribriform plate), where the mucosa was attached only by the remaining fila olfactoria. After transection and tearing of the fila olfactoria, the mucosa was harvested in one or a few pieces. Intracranial samples of the olfactory bulbs and adjacent brain regions of the frontal lobe were collected via an endoscopic endonasal transcribriform approach. The procedure was described in detail in Khan et al., see also Movie S1 in that paper. Briefly, after opening the cribriform plate with a hammer and chisel to avoid aerosol formation, the dura mater was exposed and incised longitudinally with the sickle knife. The olfactory bulbs were then dissected from the surrounding tissue (such as arachnoidea and blood vessels) and removed in an atraumatic way (blunt dissection with a ball probe) and cut posteriorly from the olfactory tract. Several biopsies of the adjacent brain regions were taken with an upbiting forceps and fixed in 10% formalin for cryopreservation and paraffin embedding. Brain samples were taken from the gyrus rectus or gyri orbitales from the frontal lobe and contained macroscopically gray and white matter. At each participating hospital, procedures were carried out by a team of two ear, nose, and throat surgeons or trainees specialized in endoscopic sinus surgery, wearing personal protective equipment: gown, gloves, and powered air purifying respirator.
A dedicated nasopharyngeal swab was taken preprocedurally and stored at −80°C. Later qRT-PCR analysis was performed by the National Reference Center for Respiratory Pathogens at the Department of Laboratory Medicine, University Hospitals Leuven (Leuven, Belgium). Viral RNA extraction was performed with the MagMAX Viral/Pathogen II kit (Thermo Fisher Scientific, Cat#A48383) on a KingFisher Flex System, followed by qRT-PCR with the TaqPath COVID-19 qRT-PCR kit (Thermo Fisher Scientific, Cat#A48067) on a QuantStudio 7 Flex platform (Thermo Fisher Scientific).
A combination of methods was used to determine the variant of concern based on RNA extracted from postmortem samples or viral cultures. 1) Whole-genome sequencing of RNA extracted from a postmortem nasopharyngeal swab by the Laboratory of Clinical and Epidemiological Virology of the KU Leuven Rega Institute (Leuven, Belgium). 2) Whole-genome sequencing of RNA extracted from a piece of formalin-fixed tissue of the inferior turbinate by GenXPro (Frankfurt, Germany). 3) Whole-genome sequencing of RNA extracted from an apical wash of HAEC-ALI cultures by GenXPro (Frankfurt, Germany). 4) Variant-specific qRT-PCR assays VoXcreen-DO, VoXcreen-BA.1-2, and VoXcreen-BA.2-4-5 on RNA extracted from a piece of formalin-fixed tissue of the inferior turbinate by GenXPro (Frankfurt, Germany). We were unable to determine the variant of concern for COVID #113 and #115. Both patients had a negative rapid antigen test result on a postmortem nasopharyngeal swab. As these cases date from well within the dominance period of Omicron in Belgium, COVID #113 and #115 are grouped in the Omicron subcohort in Figure 2.
Antibody serum titers were measured in the Department of Laboratory Medicine, University Hospitals Leuven (Leuven, Belgium) on the Abbott Architect platform with the SARS-CoV-2 IgG (anti-N) and IgG II Quant (anti-S) assays using the manufacturer’s cut-offs for positivity of 1.4 S/CO and 50 AU/mL, respectively. Serology for anti-N IgG is scored as positive vs. negative and detection of anti-N IgG is therefore referred to as "seroconversion".
Tissue samples from the 45 COVID-19 cases and 8 control cases were transferred into containers with 10% neutral buffered formalin (Sigma-Aldrich, Cat#HT5011) for >72 h to fix the tissues and inactivate SARS-CoV-2. Samples were treated for cryoprotection by immersing serially in 15%, 25%, and 30% sucrose (Sigma-Aldrich, Cat#S0389-1KG) in 1 x PBS over a period of 9–14 days. The orientation of the samples was recorded before embedding in Tissue-Tek O.C.T. compound (Sakura, Cat#4583) on dry ice. Cryosections of 6–8 μm thickness were cut on a Leica CM3050 S cryostat and collected on SuperFrost Plus Gold slides (Thermo Fisher Scientific/Menzel Gläser, Cat#K5800AMNZ72) or Superfrost Plus Micro slides (VWR, Cat#48311-703). Slides were air-dried at room temperature, and boxes of slides were sealed prior to storage at −80°C. In parallel, tissue samples of olfactory bulb and frontal lobe were processed for paraffin embedding using a fully automated platform, 5 μm paraffin-embedded sections were air-dried and stored at 4°C.
The fluorescence RNAscope platform was used to visualize viral RNA in the 45 COVID-19 cases and the 8 control cases. Most slides contained multiple sections. Staining was performed with the RNAscope manual assay using the Multiplex Fluorescent Detection Kit v2 (Advanced Cell Diagnostics, Cat#323110) according to manufacturer’s protocols. Briefly, slides were dried at 55°C overnight, then pretreated with hydrogen peroxide, followed by permeabilization in target retrieval reagent (Advanced Cell Diagnostics, Cat#322000) for 3 min in a steamer, and digestion with Protease III (Advanced Cell Diagnostics, Cat#322337) at 40°C for 15 min for cryosections and with Protease Plus (Advanced Cell Diagnostics, Cat#322330) at 40°C for 20 min for paraffin embedded sections. A combination of probes for target RNA detection was hybridized at 40°C for 2 h. Probes in the C4 channel were developed with the RNAscope 4-Plex Ancillary Kit (Advanced Cell Diagnostics, Cat#323120). Signal amplification was followed by development of appropriate HRP channels with dyes Opal 520 (Akoya Biosciences, Cat#FP1487001KT), Opal 570 (Akoya Biosciences, Cat#FP1488001KT), and Opal 690 (Akoya Biosciences, Cat#FP1497001KT). DAPI (Thermo Fisher Scientific, Cat#D1306) served as nuclear stain. Slides were mounted in Mount Solid antifade (abberior, Cat#MM-2011-2X15ML). Confocal images were taken with the Zeiss ZEN 2.6 system on a Zeiss LSM 800.
For codetection of RNA and protein, IHC was performed after the final step of HRP blocker application in the RNAscope Multiplex Fluorescent Detection protocol. Slides were blocked in 10% donkey serum (Sigma-Aldrich, Cat#S30-100ML) in 0.1% Triton/PBS at room temperature for 1 h. The following primary antibodies were diluted in 2% donkey serum in 0.1% Triton/PBS and incubated at 4°C overnight: Cytokeratin 8 (Novus Biologicals, Cat#NBP2-67468) at 1:300, MUC5AC (Thermo Fisher Scientific, Cat#MA5-12178) at 1:400, SARS-CoV-2 Nucleocapsid (Sino Biological, Cat#40143-R001) at 1:100, Somatostatin receptor subtype 2A/SSTR2A (Biotrend, Cat#NB-49-016-50ul) at 1:4000, TuJ1/TUBB3 (BioLegend, Cat#801202) at 1:100 for olfactory mucosa sections and 1:400 for olfactory bulb sections, GFAP (Novus Biologicals, Cat#NB300-141) at 1:600, NGFR p75 (Sigma-Aldrich, Cat#N5408) at 1:200, S100B (Enzo Life Sciences, Cat#ENZ-ABS307) at 1:300, Aquaporin 4 (Millipore, Cat#AB3594) at 1:100, CD31 (Abcam, Cat#ab28364) at 1:50, ZO-1 tight junction protein (Novus Biologicals, Cat#NBP1-85046) at 1:100 and OMP (FUJIFILM Wako Shibayagi, Cat#544-10001-WAKO) at 1:500. Slides were then washed in 0.1% Triton/PBS 3 × 5 min each followed by incubation with appropriate secondary antibodies at 1:500 in 2% normal donkey serum in 0.1% Triton/PBS at room temperature for 1 h. Secondary antibodies were Alexa Fluor Plus 488 donkey anti-rabbit (Thermo Fisher Scientific, Cat#A32790), Alexa Fluor Plus 555 donkey anti-rabbit (Thermo Fisher Scientific, Cat#A32794), Alexa Fluor Plus 555 donkey anti-mouse (Thermo Fisher Scientific, Cat#A32773), Alexa Fluor Plus 555 donkey anti-goat (Thermo Fisher Scientific, Cat#A32816), Alexa Fluor Plus 647 donkey anti-rabbit (Thermo Fisher Scientific, Cat#A32795), Alexa Fluor Plus 647 donkey anti-mouse (Thermo Fisher Scientific, Cat#A32787), and Alexa Fluor Plus 647 donkey anti-goat (Thermo Fisher Scientific, Cat#A32849). Slides were washed in 0.1% Triton/PBS 3 × 5 min each followed by DAPI (Thermo Fisher Scientific, Cat#D1306) application for nuclei staining. Slides were mounted in Mount Solid antifade (abberior, Cat#MM-2011-2X15ML). For IHC only, slides were pretreated in target retrieval reagent (Advanced Cell Diagnostics, Cat#322000) for 3 min in a steamer. Primary antibody application, secondary antibody detection, DAPI staining, and mounting were performed as above. Confocal images were taken with the Zeiss ZEN 2.6 system on a Zeiss LSM 800.
Human airway epithelial cells (HAEC) from healthy donors were of bronchial origin in Figures 7B and 7C (Epithelix, Cat#EP01MD) and Figure 8E (developed in-house), and of nasal origin (Epithelix, Cat#EP02MP) in Figure 8D. After arrival in the laboratory, the air-liquid interface (ALI) inserts were washed with and maintained in pre-warmed MucilAir medium (Epithelix, Cat#EP04MM) at 37°C and 5% CO2 for at least four days before use. In case of in-house developed inserts, six-week differentiated inserts in the ALI system were used that showed proper cilia beating and transepithelial electrical resistance measurements. Potential infectious particles from fresh or frozen nasopharyngeal swabs were extracted in 1 mL of MucilAir medium. Viral inocula were prepared in either MucilAir medium or control CSF sample to a final dilution of 5×103 TCID50/mL. On the day of the experiment, the HAEC-ALI cultures were exposed to 200 μL of viral inocula (equal to a multiplicity of infection of 103 TCID50/insert) or 200 μL pure CSF samples or nasopharyngeal swab solutions or fresh media at 35°C for 3 h. The apical sides of the HAEC-ALI cultures were washed with assay medium at indicated days, RNA was extracted from the wash fluids using the NucleoSpin RNA virus kit (Macherey-Nagel, Cat#740956.50), and SARS-CoV-2 RNA was quantified by qRT-PCR using the iTaq universal probes one-step kit (Bio-Rad, Cat#1725141) with a commercial mix of primers for the N gene (IDT Technologies, Cat#10006606) on a LightCycler 96 platform (Roche). The SARS-CoV-2 RNA concentrations are expressed as copies/mL in the wash fluid of the HAEC-ALI cultures by conversion from the Ct values based on a standard curve of DNA in water. The Lower-Limit-of-Quantification (LLOQ) was determined based on a dilution series of a SARS-CoV-2 virus stock treated in the same way as the samples; the LLOQ is the lowest SARS-CoV-2 concentration that was still within the linear range of this serial dilution. In Figure 8D, the inserts inoculated with the nasopharyngeal swab from COVID #108 showed fungal infection on day 10 and were not further analyzed; one insert inoculated with the nasopharyngeal swab from COVID #103 showed fungal infection on day 2 and was not further analyzed. In Figures 8C–8E, samples were tested in duplicate inserts; each replicate and the average are shown. |
|
PMC9647030 | Xuping Xie,Jing Zou,Chaitanya Kurhade,Mingru Liu,Ping Ren,Pei-Yong Shi | Neutralization of SARS-CoV-2 Omicron sublineages by 4 doses of the original mRNA vaccine | 10-11-2022 | SARS-CoV-2,neutralization,mRNA vaccine,variants,breakthrough | Since the initial emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Omicron BA.1, several Omicron sublineages have emerged, leading to BA.5 as the current dominant sublineage. Here, we report the neutralization of different Omicron sublineages by human sera collected from individuals who had distinct mRNA vaccination and/or BA.1 infection. Four-dose-vaccine sera neutralize the original USA-WA1/2020, Omicron BA.1, BA.2, BA.2.12.1, BA.3, and BA.4/5 viruses with geometric mean titers (GMTs) of 1,554, 357, 236, 236, 165, and 95, respectively; two-dose-vaccine-plus-BA.1-infection sera exhibit GMTs of 2,114, 1,705, 730, 961, 813, and 274, respectively; and three-dose-vaccine-plus-BA.1-infection sera show GMTs of 2,962, 2,038, 983, 1,190, 1,019, and 297, respectively. Thus, the four-dose vaccine elicits the lowest neutralization against BA.5; the two-dose vaccine plus BA.1 infection elicits significantly higher GMTs against Omicron sublineages than the four-dose-vaccine; and the three-dose vaccine plus BA.1 infection elicits slightly higher GMTs (statistically insignificant) than the two-dose vaccine plus BA.1 infection. Finally, the BA.2.75 is more susceptible than BA.5 to four-dose-vaccine-elicited neutralization and three-dose-vaccine-plus-BA.1-infection-elicited neutralization. | Neutralization of SARS-CoV-2 Omicron sublineages by 4 doses of the original mRNA vaccine
Since the initial emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Omicron BA.1, several Omicron sublineages have emerged, leading to BA.5 as the current dominant sublineage. Here, we report the neutralization of different Omicron sublineages by human sera collected from individuals who had distinct mRNA vaccination and/or BA.1 infection. Four-dose-vaccine sera neutralize the original USA-WA1/2020, Omicron BA.1, BA.2, BA.2.12.1, BA.3, and BA.4/5 viruses with geometric mean titers (GMTs) of 1,554, 357, 236, 236, 165, and 95, respectively; two-dose-vaccine-plus-BA.1-infection sera exhibit GMTs of 2,114, 1,705, 730, 961, 813, and 274, respectively; and three-dose-vaccine-plus-BA.1-infection sera show GMTs of 2,962, 2,038, 983, 1,190, 1,019, and 297, respectively. Thus, the four-dose vaccine elicits the lowest neutralization against BA.5; the two-dose vaccine plus BA.1 infection elicits significantly higher GMTs against Omicron sublineages than the four-dose-vaccine; and the three-dose vaccine plus BA.1 infection elicits slightly higher GMTs (statistically insignificant) than the two-dose vaccine plus BA.1 infection. Finally, the BA.2.75 is more susceptible than BA.5 to four-dose-vaccine-elicited neutralization and three-dose-vaccine-plus-BA.1-infection-elicited neutralization.
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Omicron variant was initially identified in South Africa in November 2021. Due to Omicron’s improved viral transmission and immune evasion, , the World Health Organization designated it as the fifth variant of concern (VOC) after the previous Alpha, Beta, Gamma, and Delta VOCs. Since then, the Omicron variant has evolved to several sublineages, including BA.1, BA.2, BA.2.12.1, BA.3, BA.4, and BA.5. Among these sublineages, only Omicron BA.3 remained at a low frequency in circulation, most likely due to its low fitness, whereas other sublineages sequentially increased their prevalence over time. As of September 3, 2022, in the United States, sublineages BA.2.12.1, BA.4, and BA.5 accounted for 0.1%, 2.2%, and 87.5% of the total COVID-19 cases, respectively (https://covid.cdc.gov/covid-data-tracker/#variant-proportions). Besides the above Omicron sublineages, sublineage BA.2.75 emerged in late May 2022 and has increased its prevalence in many countries. It is thus important to determine the neutralization susceptibility of the ongoing Omicron sublineages, particularly the most prevalent, BA.5, to vaccination and previous infections. Immediately after the emergence of the Omicron variant, BA.1 was found to evade vaccine-elicited neutralization more efficiently than any previous VOCs. , , , , , Two doses of mRNA vaccine did not elicit robust neutralization against BA.1; three doses of vaccine were required to generate sufficient neutralization against BA.1. Non-Omicron infection did not elicit robust neutralization against BA.1 either. Among all the currently known Omicron sublineages, BA.5 exhibited the greatest evasion of vaccine-mediated neutralization; three doses of BNT162b2 mRNA vaccine elicited weak neutralization against BA.5. The latter result, together with the high prevalence of BA.5, underscores the urgency to examine the neutralization of BA.5 after four doses of mRNA vaccine. In addition, many people who had previously received two or three doses of vaccine contracted BA.1 breakthrough infection during the initial Omicron wave; it would be important to evaluate their antibody neutralization against BA.5. To address these key questions, we have characterized the neutralization profiles of sera, obtained from humans with distinct mRNA vaccination and/or BA.1 infection, against different Omicron sublineages.
We used a set of previously established recombinant SARS-CoV-2s to determine the serum neutralization against different Omicron sublineages. Each recombinant SARS-CoV-2 contained a complete spike gene from BA.1, BA.2, BA.2.12.1, BA.3, or BA.4/5 in the backbone of USA-WA1/2020 (a virus strain isolated in January 2020) containing an mNeonGreen (mNG) reporter, resulting in BA.1-, BA.2-, BA.2.12.1-, BA.3-, or BA.4/5-spike mNG SARS-CoV-2. BA.4 and BA.5 have an identical spike sequence and are denoted as BA.4/5. Figure 1A summarizes the amino acid mutations of the spike protein from different Omicron sublineages. An mNG gene was engineered into open reading frame 7 (ORF7) of the viral genome to enable a fluorescent focus-reduction neutralization test (FFRNT) in a high-throughput format. The insertion of mNG reporter attenuated SARS-CoV-2 replication and pathogenesis. , The FFRNT has been reliably used to measure antibody neutralization for COVID-19 vaccine research and development. , Using FFRNT, we measured the neutralization of three panels of human sera against the chimeric Omicron sublineage-spike mNG SARS-CoV-2s. The first panel consisted of 25 pairs of sera collected from individuals before and after dose four of Pfizer or Moderna’s original vaccine (Table S1). Those specimens tested negative against viral nucleocapsid protein, suggesting those individuals had not been infected by SARS-CoV-2. The second and third serum panels were collected from individuals who had received two (n = 29; Table S2) or three (n = 38; Table S3) doses of the original mRNA vaccine and subsequently contracted Omicron BA.1 breakthrough infection. The BA.1 breakthrough infection was confirmed for each patient by sequencing viral RNA collected from nasopharyngeal swab samples. Tables S1–S3 summarize (1) the serum information and (2) the 50% fluorescent focus-reduction neutralization titers (FFRNT50) against USA-WA1/2020, BA.1-, BA.2-, BA.2.12.1-, BA.3-, and BA.4/5-spike SARS-CoV-2s. The description and analysis of the FFRNT50 results against different Omicron sublineages are detailed in the following sections for each serum panel.
To measure four doses of vaccine-elicited neutralization, we collected 25 pairs of sera from individuals before and after dose four of Pfizer or Moderna mRNA vaccine. For each serum pair, one sample was collected 3–8 months after dose three of the vaccine; the other sample was obtained from the same individual 1–3 months after dose four of the vaccine (Table S1). Before the fourth dose of the vaccine, the three-dose-vaccine sera neutralized USA-WA1/2020, BA.1-, BA.2-, BA.2.12.1-, BA.3-, and BA.4/5-spike mNG viruses with low geometric mean titers (GMTs) of 144, 32, 24, 25, 20, and 17, respectively (Figure 1B), and after the fourth dose of the vaccine, the GMTs increased significantly to 1,554, 357, 236, 236, 165, and 95, respectively (Figure 1C), so the fourth dose of the vaccine significantly increased the neutralization against the corresponding viruses by 10.8-, 11.2-, 9.8-, 9.4-, 8.3-, and 5.6-fold, respectively (Figure 1D). Despite the significant increase in neutralization after the fourth dose of the vaccine, the GMTs against BA.1-, BA.2-, BA.2.12.1-, BA.3-, and BA.4/5-spike viruses were 4.4-, 6.6-, 6.6-, 9.4-, and 16.4-fold lower than the GMTs against the USA-WA1/2020, respectively (Figure 1C). These results support three conclusions. First, among the tested Omicron sublineages, BA.5 possesses the greatest evasion of vaccine-elicited neutralization. Our results are in agreement with other studies supporting that BA.5 and other Omicron sublineages efficiently evade vaccine-elicited neutralization. , , , , Second, the booster effect by the fourth dose is less pronounced against BA.4/5 compared with USA-WA1/2020 and other Omicron sublineages. It should be noted that dose four did increase the neutralizing GMTs against BA.4/5 from 17 (Figure 1B) to 95 (Figure 1C). A recent study reported a neutralizing titer of 70 as the threshold to prevent breakthrough infections of the Delta variant. Although the minimal neutralizing titer required to prevent BA.5 infection has not been determined, the low neutralization against BA.5 after dose three of the vaccine (GMT of 103 at 1 month post-dose three, reported by Kurhade et al.) and dose four of the vaccine (GMT of 95 at 1 to 3 months post-dose four, reported here), together with the increased viral transmissibility, could account for the ongoing surge of BA.5 around the world. Third, an updated vaccine that matches the highly immune-evasive and prevalent BA.5 spike is needed to mitigate the current and future Omicron surges. Our results support the US Food and Drug Administration’s recommendation to include BA.5 spike for future COVID-19 vaccine booster doses.
To compare with four-dose-vaccine sera, we measured the neutralization against Omicron sublineages using sera collected from individuals who had received two or three doses of the original mRNA vaccine and subsequently contracted BA.1 infection (Figure 2 ). Tables S2 and S3 summarize the FFRNT50 results for two-dose-vaccine-plus-BA.1-infection sera and three-dose-vaccine-plus-BA.1-infection sera, respectively. The two-dose-vaccine-plus-BA.1-infection sera neutralized BA.1, BA.2, BA.2.12.1, BA.3, and BA.4/5 with GMTs of 2,114, 1,705, 730, 961, 813, and 274, respectively (Figure 2A), and the three-dose-vaccine-plus-BA.1-infection sera showed slightly higher GMTs of 2,962, 2,038, 983, 1,190, 1,019, and 297, respectively (Figure 2B). So, the GMT ratios between the three-dose-vaccine-plus-BA.1-infection sera and the two-dose-vaccine-plus-BA.1-infection sera were 1.4, 1.2, 1.3, 1.2, 1.3, and 1.1 when neutralizing USA-WA1/2020, BA.1-, BA.2-, BA.2.12.1-, BA.3-, and BA.4/5-spike viruses, respectively; these GMT differences between the two serum groups were statistically insignificant, suggesting that the extra dose of vaccine does not significantly boost neutralization for the three-dose-vaccine-plus-BA.1-infection sera. In contrast, the GMT ratios between the two-dose-vaccine-plus-BA.1-infection and the four-dose-vaccine sera were 1.4, 4.8, 3.1, 4.1, 4.9, and 3.9 when neutralizing USA-WA1/2020, BA.1-, BA.2-, BA.2.12.1-, BA.3-, and BA.4/5-spike viruses, respectively. The result suggests that, compared with the two extra doses of vaccine in the four-dose-vaccine sera, the BA.1 infection in the two-dose-vaccine-plus-BA.1-infection sera is more efficient in boosting both the magnitude and breadth of neutralization against all Omicron sublineages; however, the neutralization against BA.5 was still the lowest among all tested sublineages. For the two-dose-vaccine-plus-BA.1-infection sera, the GMTs against BA.1-, BA.2-, BA.2.12.1-, BA.3-, and BA.4/5-spike viruses were 1.2-, 2.9-, 2.2-, 2.6-, and 7.7-fold lower than the GMT against the USA-WA1/2020, respectively (Figure 2A); similar results were observed for the three-dose-vaccine-plus-BA.1-infection sera, with GMTs against BA.1-, BA.2-, BA.2.12.1-, BA.3-, and BA.4/5-spike viruses that were 1.5-, 3.0-, 2.5-, 2.9-, and 10-fold lower than the GMT against the USA-WA1/2020, respectively (Figure 2B). The GMT decreases against Omicron sublineages for the two-dose-vaccine-plus-BA.1-infection sera and those for the three-dose-vaccine-plus-BA.1-infection sera are significantly less than those observed for the four-dose-vaccine sera (compare Figures 2A and 2B with Figure 1C). The results again indicate that BA.1 infection of vaccinated people efficiently boosts the breadth of neutralization against all tested Omicron sublineages. However, such BA.1-infection-mediated boost of neutralizing magnitude/breadth is dependent on previous vaccination. This is because BA.1 infection of unvaccinated people did not elicit greater neutralizing magnitude/breadth against Omicron sublineages than three doses of mRNA vaccine.
To assess the neutralization of sublineage BA.2.75, we engineered the complete spike gene of BA.2.75 (Figure 1A) into the backbone of mNG USA-WA1/2020, resulting in BA.2.75-spike mNG SARS-CoV-2. The BA.2.75-spike mNG SARS-CoV-2 was sequenced to ensure no undesired mutations. When tested with four-dose-vaccine sera, the neutralizing GMT against BA.2.75-spike virus was 2.8-fold higher than that against BA.5-spike virus (Figure 1C). Similarly, when tested with three-dose-vaccine-plus-BA.1-infection sera, the neutralizing GMT against BA.2.75-spike virus was 3.4-fold higher than that against BA.5-spike virus (Figure 2B). Collectively, the results indicate that BA.2.75 is less immune-evasive than BA.4/5.
Our study has several limitations. First, this study lacks analysis of sera from vaccinated people who were infected with Omicron sublineages other than BA.1. It would be particularly important to study vaccinated individuals who contracted BA.5 infection. Second, we have not analyzed T cells and non-neutralizing antibodies that have Fc-mediated effector functions. These two immune arms, together with neutralizing antibodies, protect patients from severe disease. Many T cell epitopes after vaccination or natural infection are preserved in Omicron spikes. Third, the relatively small sample size and heterogeneity in vaccination/infection time have weakened our ability to compare the results among the three distinct serum panels. Regardless of these limitations, our results consistently showed that (1) the booster effect by dose four of the mRNA vaccine is less pronounced against BA.4/5 compared with USA-WA1/2020 and other Omicron sublineages; (2) four doses of the current mRNA vaccine elicit low neutralization against BA.5 spike, rationalizing the need to update the vaccine sequence to match the highly immune-evasive and prevalent BA.5; and (3) vaccinated individuals with BA.1 breakthrough infection develop greater neutralizing magnitude/breadth against Omicron sublineages than those who have received four doses of mRNA vaccine. These laboratory investigations, together with real-world vaccine effectiveness data, will continue to guide vaccine strategy and public health policy.
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Pei-Yong Shi ([email protected]).
The Omicron-spike SARS-CoV-2 has been deposited to the World Reference Center for Emerging Viruses and Arboviruses (https://ww.utmb.edu/wrceva) at UTMB for distribution. All reagents generated in this study are available from the lead contact with a completed Materials Transfer Agreement.
• All data reported in this paper will be shared by the lead contact upon request. • This paper does not report original code. • Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.
All virus work was performed in a biosafety level 3 (BSL-3) laboratory with redundant fans in the biosafety cabinets at The University of Texas Medical Branch at Galveston. All personnel wore powered air-purifying respirators (Breathe Easy, 3M) with Tyvek suits, aprons, booties, and double gloves. The research protocol regarding the use of human serum specimens was reviewed and approved by the University of Texas Medical Branch (UTMB) Institutional Review Board (IRB number 20–0070). No informed consent was required since these deidentified sera were leftover specimens before being discarded. No diagnosis or treatment was involved either.
Vero E6 (ATCC® CRL-1586) was purchased from the American Type Culture Collection (ATCC, Bethesda, MD), and maintained in a high-glucose Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS; HyClone Laboratories, South Logan, UT) and 1% penicillin/streptomycin at 37°C with 5% CO2. Culture media and antibiotics were purchased from ThermoFisher Scientific (Waltham, MA). The cell line was tested negative for mycoplasma.
Viruses were recovered from infectious cDNA clones of SARS-CoV-2 and propagated on the Vero-E6 TMPRSS2 cells in the DMEM medium (GIBCO) supplemented with 2% fetal bovine serum (FBS) and 1%penecillin/streptomycin (GIBCO).
Three panels of human sera were used in the study. The first panel consisted of 25 pairs of sera collected from individuals 3–8 months after vaccine dose 3, and no more than 3 months after dose 4 of the Pfizer-BioNTech or Moderna vaccine. This panel had been tested negative for SARS-CoV-2 nucleocapsid protein expression using Bio-Plex Pro Human IgG SARS-CoV-2 N/RBD/S1/S2 4-Plex Panel (Bio-rad). The second serum panel (n = 29) was collected from individuals who had received 2 doses of mRNA vaccine and subsequently contracted Omicron BA.1. The third serum panel (n = 38) was collected from individuals who had received 3 doses of mRNA vaccine and subsequently contracted Omicron BA.1. The genotype of infecting virus was verified by the molecular tests with FDA’s Emergency Use Authorization and Sanger sequencing. The de-identified human sera were heat-inactivated at 56°C for 30 min before the neutralization test. The serum information is presented in Tables S1–S3. The age distribution of the human patients in each panel is the following: panel 1 aged 59–94 yrs with a medium of 80 yrs, panel 2 aged 18–96 yrs with a media of 46 yrs, panel 3 aged 26–84 yrs with a medium of 64 yrs. Both genders of human patients were included. The gender distribution of the human patients in each panel is the following: 52%/48% (female/male) in panel 1, 55%/45% in panel 2; 55%/45% in panel 3.
Recombinant Omicron sublineage BA.1-, BA.2-, BA.2.12.1-, BA.3-, BA.4/5-spike mNG SARS-CoV-2s that was constructed by engineering the complete spike gene from the indicated variants into an infectious cDNA clone of mNG USA-WA1/2020 were reported previously. , BA.2.75-spike sequence was based on GISAID EPI_ISL_13521499. Figure 1A depicts the spike mutations from different Omicron sublineages. The full-length cDNA of the viral genome bearing the variant spike was assembled via in vitro ligation and used as a template for in vitro transcription. The full-length viral RNA was then electroporated into Vero E6-TMPRSS2 cells. On day 3–4 post electroporation, the original P0 virus was harvested from the electroporated cells and propagated for another round on Vero E6 cells to produce the P1 virus. The infectious titer of the P1 virus was quantified by fluorescent focus assay on Vero E6 cells and sequenced for the complete spike gene to ensure no undesired mutations. The P1 virus was used for the neutralization test. The protocols for the mutagenesis of mNG SARS-CoV-2 and virus production were reported previously.
A fluorescent focus reduction neutralization test (FFRNT) was performed to measure the neutralization titers of sera against USA-WA1/2020, BA.1-, BA.2-, BA.2.12.1-, BA.3-, and BA4/5-spike mNG SARS-CoV-2. The FFRNT protocol was reported previously. Vero E6 cells were seeded onto 96-well plates with 2.5×104 cells per well (Greiner Bio-one™) and incubated overnight. On the next day, each serum was 2-fold serially diluted in a culture medium and mixed with 100–150 focus-forming units of mNG SARS-CoV-2. The final serum dilution ranged from 1:20 to 1:20,480. After incubation at 37°C for 1 h, the serum-virus mixtures were loaded onto the pre-seeded Vero E6 cell monolayer in 96-well plates. After 1 h infection, the inoculum was removed and 100 μL of overlay medium containing 0.8% methylcellulose was added to each well. After incubating the plates at 37°C for 16 h, raw images of mNG foci were acquired using Cytation™ 7 (BioTek) armed with 2.5× FL Zeiss objective with a wide field of view and processed using the software settings (GFP [469,525] threshold 4000, object selection size 50–1000 μm). The fluorescent mNG foci were counted in each well and normalized to the non-serum-treated controls to calculate the relative infectivities. The FFRNT50 value was defined as the minimal serum dilution to suppress >50% of fluorescent foci. The neutralization titer of each serum was determined in duplicate assays, and the geometric mean was taken. Tables S1–S3 summarize the FFRNT50 results.
Data were plotted as scatted dots. The geometric mean with 95% confidence intervals was presented. Sample size (n represents the number of human serum samples) was indicated in the corresponding figures. The nonparametric Wilcoxon matched-pairs signed rank test was used to analyze the statistical significance in Figures 1 and 2. Friedman with Dunn’s multiple comparisons test was performed to assess the statistical significance of the increase in the neutralization of sera against variants in Figure 1D. Figures were initially plotted in GraphPad Prism and assembled using Adobe illustrator. |
|
PMC9647041 | Katherine M. Hannan,Priscilla Soo,Mei S. Wong,Justine K. Lee,Nadine Hein,Perlita Poh,Kira D. Wysoke,Tobias D. Williams,Christian Montellese,Lorey K. Smith,Sheren J. Al-Obaidi,Lorena Núñez-Villacís,Megan Pavy,Jin-Shu He,Kate M. Parsons,Karagh E. Loring,Tess Morrison,Jeannine Diesch,Gaetan Burgio,Rita Ferreira,Zhi-Ping Feng,Cathryn M. Gould,Piyush B. Madhamshettiwar,Johan Flygare,Thomas J. Gonda,Kaylene J. Simpson,Ulrike Kutay,Richard B. Pearson,Christoph Engel,Nicholas J. Watkins,Ross D. Hannan,Amee J. George | Nuclear stabilization of p53 requires a functional nucleolar surveillance pathway | 01-11-2022 | nucleolar surveillance pathway,nucleolus,p53,ribosome biogenesis,high-throughput screening,ribosomal proteins,stress,high-content screening | Summary The nucleolar surveillance pathway monitors nucleolar integrity and responds to nucleolar stress by mediating binding of ribosomal proteins to MDM2, resulting in p53 accumulation. Inappropriate pathway activation is implicated in the pathogenesis of ribosomopathies, while drugs selectively activating the pathway are in trials for cancer. Despite this, the molecular mechanism(s) regulating this process are poorly understood. Using genome-wide loss-of-function screens, we demonstrate the ribosome biogenesis axis as the most potent class of genes whose disruption stabilizes p53. Mechanistically, we identify genes critical for regulation of this pathway, including HEATR3. By selectively disabling the nucleolar surveillance pathway, we demonstrate that it is essential for the ability of all nuclear-acting stresses, including DNA damage, to induce p53 accumulation. Our data support a paradigm whereby the nucleolar surveillance pathway is the central integrator of stresses that regulate nuclear p53 abundance, ensuring that ribosome biogenesis is hardwired to cellular proliferative capacity. | Nuclear stabilization of p53 requires a functional nucleolar surveillance pathway
The nucleolar surveillance pathway monitors nucleolar integrity and responds to nucleolar stress by mediating binding of ribosomal proteins to MDM2, resulting in p53 accumulation. Inappropriate pathway activation is implicated in the pathogenesis of ribosomopathies, while drugs selectively activating the pathway are in trials for cancer. Despite this, the molecular mechanism(s) regulating this process are poorly understood. Using genome-wide loss-of-function screens, we demonstrate the ribosome biogenesis axis as the most potent class of genes whose disruption stabilizes p53. Mechanistically, we identify genes critical for regulation of this pathway, including HEATR3. By selectively disabling the nucleolar surveillance pathway, we demonstrate that it is essential for the ability of all nuclear-acting stresses, including DNA damage, to induce p53 accumulation. Our data support a paradigm whereby the nucleolar surveillance pathway is the central integrator of stresses that regulate nuclear p53 abundance, ensuring that ribosome biogenesis is hardwired to cellular proliferative capacity.
Mutations in the potent tumor suppressor protein p53 and its effector pathways occur in the majority of human cancers and are therefore the subject of intense investigation. A key mechanism by which p53 is regulated is at the level of protein stabilization, through the murine double minute 2 homolog (MDM2) protein, a nuclear-localized E3 ubiquitin ligase that induces ubiquitination, and subsequently proteasomal degradation of p53. DNA damage from ionizing radiation or certain chemotherapeutic agents leads to the amino-terminal phosphorylation of p53, which prevents MDM2 binding and results in p53 stabilization. This triggers a number of anti-proliferative programs by activating or repressing key effector genes in a context-dependent manner (Zilfou and Lowe, 2009). The p53-MDM2 interaction is also antagonized by the tumor suppressor p14ARF in response to oncogenic challenges (Dai et al., 2012). More recently, a third mechanism of p53 stabilization has been identified: the nucleolar surveillance pathway (NSP), which is activated by acute disruptions to ribosome biogenesis (RiBi), resulting in inhibitory binding of certain ribosomal proteins (RPs) to MDM2, thus leading to increased abundance of nuclear p53 protein (Boulon et al., 2010; Donati et al., 2013; Sloan et al., 2013). In contrast to the former, the precise mechanisms underlying p53 stabilization in response to the NSP are poorly understood. For example, the ribosomal proteins RPL5 (uL18) and RPL11 (uL5) have been implicated as the central regulators of the NSP through their participation in the 5S ribonucleoprotein particle (5S-RNP) complex that binds to and inactivates MDM2 in response to nucleolar stress (Donati et al., 2013; Sloan et al., 2013). However, other RP and non-RP genes have also been implicated in regulating the NSP signaling process, suggesting the definitive mechanism is yet to be resolved. It is also unclear why loss or inactivation of only certain ribosome-associated genes gives rise to increased p53 stabilization or relates to ribosomopathies. Finally, the functional relationship of the NSP to the mechanisms underlying p53 stabilization observed in response to “classic” non-nucleolar stress pathways, such as proteasomal stress, hypoxia, or DNA damage, is not clear.
To better understand the molecular mechanism underlying the NSP and its role in modulating p53, we first identified the entire repertoire of genes whose deletion activates stress pathways leading to stabilization of p53 in A549 (human lung adenocarcinoma, p53 wild-type) cells, which, while hypotriploid in a small proportion of the cell population, have previously been used in the literature to study the effects of RP depletion on the p53 pathway (Bursac et al., 2012; Fumagalli et al., 2009, 2012). This was achieved using a high-throughput genome-wide RNA interference (RNAi) imaging-based screen measuring nuclear p53 accumulation using immunofluorescence (“p53 stabilization screen,” Figure 1A and Table S1). The screen “cutoff” was functionally defined as the minimum amount of p53 accumulation required to induce a significant cell-cycle defect (Figures S1A–S1D), which we identified as ∼2-fold increase in p53 protein expression. Applying this cutoff (Log2 ≥ 1) to the screening dataset, 827 genes fulfilled this criterion (defined as “p53 positive,” Figure 1B). We further interrogated the “p53 positive” candidates to identify which molecular pathways/functions were enriched in the dataset using the KEGG network enrichment analysis feature of STRING (von Mering et al., 2003) (Figure 1C and Table S2). This revealed an enrichment of six major classes of genes including ribosome, nucleolus, proteasome, RNA splicing, cell cycle, and RNA polymerase II (Pol II). These classes were also broadly confirmed by gene ontology (GO) analysis (Figure 1D and Table S3), resulting in GOs relating to RNP complex and RiBi, ribosomal RNA (rRNA) processing, and rRNA metabolic processes being among the most significantly enriched. Moreover, intersecting our “p53 positive” candidate list with the LOCATE subcellular localization database (Sprenger et al., 2008), we identified a significant over-representation of proteins localized to the nucleolus, nucleus, and centrosome and, conversely, an under-representation of proteins located within the plasma membrane (Figure 1E and Table S4). Collectively, these observations strongly support the notion that perturbations in RiBi and/or the nucleolus are a major, if not the most potent regulators of p53 accumulation. We initially focused specifically on RP genes given their prominence in the dataset; ∼80% of the RPs screened were “p53 positive” when depleted, including RP genes associated with Diamond-Blackfan anemia (DBA; e.g. RPS19, RPL35A, RPS7, RPS10, RPS24, RPS26, RPL26) (Ulirsch et al., 2018) (Figure 2A). In a complementary approach, we evaluated the RPs using a quantitative total p53 assay (Alphascreen) to verify p53 expression and observed a significant correlation between the results from both techniques (Figures 2B and S2A). In total, 77.3% of the RPs specific to the 60S and 81.3% to the 40S ribosomal subunits, when depleted, induced a “p53 positive” phenotype, implying that RPs to either subunit contributed similarly to the NSP p53 response. This finding is in contrast to a study reporting that the large subunit RPs have a more profound p53 response when depleted (Nicolas et al., 2016), though the p53 threshold was set at 5-fold increase in that study. Importantly, the differential ability of the RPs when depleted to elicit p53 stabilization was not due to the inability of the siRNA to deplete the RP mRNA and protein (Figures 2C and S2B). We also established that RPS19 depletion does not alter TP53 transcript expression, but it does lead to increased p53 protein expression and downstream expression of p53 target genes including CDKN1A (cyclin dependent kinase inhibitor 1a, also known as p21), MDM2, and BBC3 (Bcl2 binding component 3, also known as PUMA) (Figure S2C). Under RPS19 depletion conditions, we also assessed translation of p53 mRNA on the polysomes (Figure S2D) and stability of the p53 protein by treating with cycloheximide (Figure S2E). Together, these data indicate that the early increase in p53 protein is likely due to increased stabilization but is still translated (but not heavily) at a later time point, presumably to sustain the response. We further examined whether the ability of an RP to induce the NSP correlated with the degree to which its depletion affected ribosome subunit biogenesis and function. We measured the abundance of the 40S and 60S ribosomal subunits and the levels of mature ribosomes (80S) bound to mRNAs in polysomes following RP depletion. Consistent with the prediction, depletion of RPL21 (eL21), RPS18 (uS13), and RPS19 (eS19), all of which induced robust stabilization of p53, also robustly reduced the abundance of the corresponding 40S/60S subunit in which they are located, as well as the number of polysomes (Figures 2D, 2E, and S3). In contrast, depletion of RPL22 (eL22) and RPL28 (eL28), which failed to induce p53 stabilization, did not impact 60S biogenesis nor the number of polysomes compared with siNT (Figures 2D, 2E, and S3). Exceptions to this were RPL5 and RPL11, whose knockdown failed to stabilize p53, even though 60S biogenesis was ablated. This observation is consistent with studies implicating “free” RPL5 and RPL11 (i.e., not incorporated into a 60S) as essential for the NSP due to their ability to bind MDM2 as part of the 5S-RNP (Bursac et al., 2012; Fumagalli et al., 2012; Macias et al., 2010; Sloan et al., 2013). We considered whether the location of an RP within the ribosome may predict its ability to disrupt ribosome assembly, and thus mediate p53 accumulation, upon depletion. To do this, we mapped the p53 intensity resulting from the knockdown of each RP onto the structure of the 60S and 40S subunits (Khatter et al., 2015) (Figure S4). While RPS18 and RPS19 (corresponding to two of the highest p53 intensities observed in the screen) co-located in the same region within the 40S subunit, there was no other clear evidence supporting that the specific location of an RP within the ribosome would increase p53 stabilization if depleted. Finally, we tested the hypothesis that RPs that integrate early into their respective ribosomal subunit (i.e., within the nucleolus) might be essential for the core structure, thus when depleted, would have the most profound effect on ribosome assembly and the NSP. By comparing p53 intensity and the published timing of integration of each RP into the ribosome (de la Cruz et al., 2015) (Table S5), we demonstrated that the p53 levels were significantly higher following knockdown of those RPs that integrate into their respective subunits during early nucleolar stages of ribosome assembly (Figure 2F). Thus, the ability of RPs to stabilize p53 correlated with their propensity to cause significant disruption to ribosome subunit assembly when depleted. This may explain, at least in part, why not all components of the ribosome, when mutated or deleted, contribute to ribosomopathies. For example, RPL22 and RPL28, which do not perturb ribosome subunit assembly when depleted, have not been associated with DBA to date. Having identified the major classes of genes, including RPs, whose deletion leads to stabilization of p53, we next determined the role of the NSP in this process; a priori, we predicted that only those genes directly involved in RiBi would be dependent on the NSP to stabilize p53 when depleted. To address this question, we took an unbiased approach to identify the key components of NSP that can be targeted to inactivate NSP-mediated p53 stabilization. Accordingly, we performed a genome-wide RNAi screen to determine the genes whose depletion suppressed p53 accumulation in response to nucleolar stress induced by knockdown of RPS19, the prototypical DBA gene known to induce NSP when depleted (Jaako et al., 2011, 2015; Sjogren et al., 2015) (termed “modifiers of ribosomal stress” screen; Figure 3A and Table S6). Using a cutoff for normalized p53 intensity of up to and including 0.5 (Log2 = −1, calculated based on 3 standard deviations (SD) above the positive control, siTP53 + siRPS19), we identified 64 genes essential for a functional NSP (Figures 3A and S5A). We rescreened these 64 candidates (outlined in STAR Methods), to identify candidates that recapitulated the primary screen phenotype (i.e., suppressed p53 response when co-depleted with siRPS19) with two or more siRNA duplexes. Critically, in addition to TP53, both RPL5 and RPL11 were the top ranked candidates that, when depleted, reduced p53 accumulation in response to NSP activation, while no other RPs reached this cutoff. This observation is in contrast to previous reports suggesting a variety of RPs regulate p53 stability, e.g., RPL23 (uL14) (Dai et al., 2004; Jin et al., 2004), RPL26 (uL24) (Ofir-Rosenfeld et al., 2008; Zhang et al., 2010), RPS3 (uS3) (Yadavilli et al., 2009), RPS7 (eS7) (Chen et al., 2007), RPS14 (uS11) (Zhou et al., 2013), RPS25 (eS25) (Zhang et al., 2013b), RPS27A (eS31) (Sun et al., 2011), RPS27 (eS27) and RPS27L (Xiong et al., 2011), RPS15 (uS19), RPS20 (uS10), and RPL37 (eL37) (Daftuar et al., 2013). Our study, therefore, functionally defines RPL5 and RPL11 as the only RPs essential for the NSP, consistent with their proposed role in the 5S-RNP interaction with MDM2. Similarly, non-RP factors previously reported to be linked to RiBi and p53 activity, e.g., SRSF1, GLTSCR2 (PICT1), HEXIM1, MYBBP1A, RRP8 (NML), and NPM1 (Fregoso et al., 2013; Kumazawa et al., 2011; Kuroda et al., 2011; Lew et al., 2012; Lindstrom, 2011; Sasaki et al., 2011), were not identified as high-ranking candidates, suggesting, at least in this system under the kinetics used for the assays, they are not essential for NSP-induced stabilization of p53 and/or may play tissue-specific or developmentally specific roles in the NSP. In addition to TP53, RPL5, and RPL11, we further validated a selection of candidates from the screen including HEATR3, RXRA, and CIRH1A as bone fide modulators of the p53 response in A549 cells and further validated using the MCF10A (mammary epithelial) cell line (Figures 3B, S5A, S5B, S5C–S5H, S5I–S5N, and S6). HEATR3 (HEAT-repeat containing 3) was of significant interest as a novel direct regulator of the NSP and the 5S-RNP-MDM2 axis, as bioinformatic domain alignment suggested that HEATR3 is a human homolog of yeast symportin 1 (Syo1) protein, which enables import of rpL5 and rpL11 into the nucleus of Saccharomyces cerevisiae (Calvino et al., 2015; Kressler et al., 2012), and acts as a scaffold for 5S-RNP biogenesis prior to incorporation into the pre-60S ribosomal subunit (Calvino et al., 2015). To analyze any structural similarities between the human and yeast proteins, we modeled the HEATR3 structure based on the Chaetomium thermophilium Syo1 (ctSyo1) crystal structure (Kressler et al., 2012) using “Modeller” (Figure 3C), which indicates the potential for RPL5 and RPL11 binding on opposite sides of the HEAT repeats similar to that shown for ctSyo1 (Calvino et al., 2015; Kressler et al., 2012). In support of this model, co-immunoprecipitation experiments from A549 cells co-transfected with myc-tagged HEATR3 (MT-HEATR3) and either FLAG-tagged RPL5 or RPL11 confirmed that RPL11 and RPL5 bind to HEATR3 in situ (Figure 3D). Moreover, depletion of HEATR3 partially phenocopied RPL5 and RPL11 knockdown, resulting in a marked reduction in 60S subunit production (Figures 3E and 3F), number of polysomes (Figure S3), 28S rRNA levels (Figures 3G, 3H, S7A–S7C, and S7D–S7G) and 5S-RNP binding to MDM2 (Figures 3I and 3J). We also assessed whether HEATR3, RPL5, and RPL11 regulate each other by knocking down each individual target and expressing expression of each protein (Figure S7H). While we observed that HEATR3 and RPL5 (but not RPL11) may regulate each other’s expression at the protein level, the reduced efficacy of HEATR3 depletion to disrupt RiBi and NSP compared with RPL5 and RPL11 suggests there may also be HEATR3-independent pathways by which RPL5 and RPL11 can assemble into the 5S-RNP. Even so, in toto, these findings suggest that HEATR3 is a functional homolog of Syo1 and important for 60S assembly and NSP in human cells through its ability to interact with the 5S-RNP (Figure 3K). Further studies are required to definitively demonstrate that RPL5, RPL11, and HEATR3 co-exist in a complex together. Having functionally defined RPL5, RPL11, and HEATR3 as direct regulators of the NSP, we next used their depletion (RNAi) to determine how important a functional NSP is for stabilization of p53 by stresses not traditionally implicated in RiBi. To do this, a representative selection of the 827 genes identified as “p53 positive” (i.e., whose depletion increased p53 levels; Figures 1B and 1C; 232 genes representing nucleolar, ribosome, splicing, Pol II, proteasome, cell cycle, and other gene classes) were rescreened to determine if their ability to stabilize p53 when depleted was dependent on the NSP (Figure 4A and Table S7). As expected, the ability of RP and other nucleolar/RiBi-related genes to robustly activate p53 when depleted was blocked when the NSP was inactivated by co-depletion of either RPL5, RPL11, or HEATR3. Notably, the overall effect of HEATR3 depletion to reduce p53 accumulation was less profound than RPL5 and RPL11 depletion, and for a subset of large RPs, HEATR3 depletion completely failed to block induction of p53 (Figures 4A and 4B and Table S7). Thus, HEATR3 is necessary for 5S-RNP-MDM2 complex assembly in response to the disruption of many (but not all) RiBi proteins, consistent with the observations above that HEATR3-independent pathways by which RPL5 and RPL11 can assemble into 5S-RNP may exist. Critically, and unexpectedly, the ability of major classes of genes not traditionally associated with the ribosome or the nucleolus (e.g., RNA splicing, cell cycle, and Pol II; Figure 1C) to stabilize p53 following their depletion was also ablated upon co-knockdown of RPL5 or RPL11, and to a lesser degree HEATR3 (Figures 4A and 4B and Table S7). The p53 suppression was not simply due to reduced ribosome assembly (and therefore reduced p53 mRNA translation) as a consequence of RPL5 or RPL11 depletion (Figures 3E, 3F, and S3), because co-depletion of RPS19 failed to blunt p53 accumulation, despite RPS19 depletion causing a similar defect in ribosomal subunit assembly and polysomes (Figures 2D, 2E, and S3). Together, these data suggest the NSP is required for robust stabilization of p53 in response to the dysregulation of a large number of eukaryotic genes and cellular processes that are not traditionally associated with RiBi. Given these unexpected findings, we extended these studies to determine the requirement of a functional NSP to mediate stabilization of p53 in response to a broad range of pharmacological agents and pathophysiologic stressors, including inhibitors of Pol I and II, nucleic acid synthesis inhibitors, agents that induce DNA damage, nuclear export inhibitors, and proteotoxic stress. Intriguingly, inactivation of the NSP by either RPL5 or RPL11 depletion ablated the ability of Pol I/II inhibitors, nucleic acid synthesis inhibitors, and all classes of DNA damage-inducing agents to stabilize p53. In contrast, the ability of proteotoxic stresses including proteasomal inhibitors, nuclear transport inhibitors, and heat shock to increase p53 levels was only moderately, or not at all blunted by inactivation of the NSP (Figures 4C and S8A). We also confirmed these findings using a high-content screening-based approach (Figure S8B), where HEATR3 depletion also blunted the response, however, not as efficiently as RPL5/L11 depletion. We noted that knockdown of RPL5 was consistently more potent at blocking the NSP compared with RPL11 or HEATR3, suggesting RPL5 may modulate p53 by mechanisms in addition to inhibitory binding of 5S-RNP to MDM2. Consistent with this, we observed that knockdown of RPL5 but not RPL11 significantly reduced p53 mRNA levels (Figure S8C), although the mechanism of this reduction was not investigated further. To further validate our results in a model of NSP inactivation (other than RPL5 and RPL11 depletion), we used embryonic fibroblasts (MEFs) isolated from mice harboring the Mdm2C305F mutation, which disrupts RPL5 and RPL11 (ergo 5S-RNP) binding to MDM2, thereby inactivating the NSP (Macias et al., 2010; Sloan et al., 2013). We again tested a complement of nuclear and physiological stressors (Figures 4D and S8D) in these cells. Consistent with RPL5/RPL11 knockdown, the Mdm2C305F mutation prevented the p53 response upon exposure to Pol I/II inhibitors, nucleic acid synthesis inhibitors, and all classes of DNA damage-inducing agents, but not proteotoxic stress. Taken together, the data indicate that an intact NSP is required for the stabilization of p53 in response to a broad range of cellular stresses, not just ribosomal/nucleolar stress. Notably, the quantitative effect of Mdm2C305F mutation to blunt p53 accumulation in response to stress more closely reflected the effect of RPL11 depletion than RPL5 depletion, consistent with the conclusions above that RPL5 may modulate p53 by mechanisms in addition to inhibitory binding of 5S-RNP to MDM2.
In summary, using global screening approaches, we have identified the complement of genes and pathways functionally required for stabilization of p53 in response to the canonical NSP. Our data definitively demonstrate that RPL5 and RPL11 do not induce p53 stabilization when depleted, and in contrast to other studies (Chen et al., 2007; Dai et al., 2004; Yadavilli et al., 2009), they are the only RPs essential for a functional NSP to stabilize p53, at least in A549 cells. Furthermore, we demonstrate that one of the top hits, HEATR3, a putative mammalian ortholog of the yeast Syo1 protein, is a ribosome assembly factor required for mammalian 60S ribosomal subunit assembly through binding of RPL5 and RPL11. Consistent with an essential role for HEATR3 in NSP-mediated stabilization of p53, HEATR3 depletion leads to reduced association of the 5S-RNP with MDM2. Intriguingly, the National Cancer Institute Genomic Data Commons Portal (https://portal.gdc.cancer.gov) demonstrates recurrent loss-of-function mutations in HEATR3 (predominantly missense) spanning the entire protein coding region; these are associated in humans with both solid and hematological malignancies and blood disorders. Thus HEATR3, like RPL5 and RPL11, may function as a tumor suppressor and a likely causative gene in the congenital disorder DBA through its role in modulating p53; this has recently been demonstrated (O'Donohue et al., 2022). Interestingly, the only previous studies addressing HEATR3 function have implicated it in NOD2-mediated NF-κB signaling (Zhang et al., 2013a), suggesting this protein may have evolved additional cellular functions outside of RiBi. Critically, by inactivating the NSP, we demonstrate that pharmacological agents and pathophysiological conditions leading to genotoxic stress, as well as the majority of genes whose loss of function stabilizes p53, do so predominantly in an NSP-dependent fashion. Our data provide experimental support to Rubbi and Milner’s original hypothesis that the nucleolus, through the NSP, is a universal stress sensor responsible for p53 homeostasis within cells (Rubbi and Milner, 2003). Thus, we conclude that the well-described mechanisms of genotoxic stress that induce extensive post-translational regulation of p53, thereby modulating its interaction with MDM2, are insufficient in the absence of a functioning NSP to robustly stabilize p53. The exception to this rule appears to be pathological conditions and pharmacologic agents that result in proteasomal stress, which stabilize p53 largely independently of the NSP. This is consistent with MDM2-mediated degradation of p53 being dependent on a functioning proteasome to degrade ubiquitinated p53. The differential ability of ribosome components to induce p53 stabilization following their depletion correlated directly with the degree of disruption of RiBi and ribosome assembly. While we acknowledge that there may be NSP-independent mechanism(s) that contribute to this observation, by extrapolation, we propose that most nuclear-acting pathological conditions, pharmacologic agents, or genetic inactivating lesions stabilize p53 in a 5S-RNP-MDM2 dependent fashion through disruption of RiBi. Consistent with this, ribosomal DNA (rDNA) is highly sensitive to DNA damage; a single lesion in the rDNA is sufficient to cause cell-cycle arrest (van Sluis and McStay, 2015), and most cytotoxic drugs and pathological conditions that induce DNA damage have been reported to cause defects in RiBi. We propose that the nucleolus functions as the cellular equivalent of a sentinel or “canary in the coal mine” to detect a broad range of cellular stresses and mediate stabilization of p53. In this model, the nucleolus acts a sensitivity gauge, whereby stresses such as DNA damage can only stabilize p53 if the stress is of sufficient magnitude to perturb RiBi/nucleolar function, thereby preventing minor cellular insults from inappropriately inhibiting proliferation. Given that RiBi is the most energy-expensive process a cell undertakes, the evolution of such a mechanism also ensures RiBi remains hardwired to proliferative capacity through p53 activity. Finally, due to the central role RiBi and the NSP play in the regulation of p53, we suggest a paradigm shift in thinking is required for how this axis contributes to cancer pathogenesis. Due to the pervasive stress to which tumor cells are exposed, we propose that overcoming NSP-induced p53 activation is likely to be a very frequent step in malignant transformation. Indeed, RP genes are hemizygously deleted in 43% of human cancers, and almost always in concert with TP53 mutations, while such RP deletions are infrequent in TP53-intact tumors (Ajore et al., 2017; Fancello et al., 2017). This is consistent with chronic activation of the NSP in response to RP deletion being incompatible with malignant transformation and negatively selected for unless p53 is inactivated.
Our study has focused on investigating the mechanism underlying the p53-dependent NSP, but it should be recognized that the NSP can also be activated independently of p53 stabilization. We also cannot rule out the possibility that p53 may also be stabilized independently of the NSP, which may also contribute (in part) to the overall response that we observed. With future advances in high-content imaging and image analysis approaches, studying simultaneous measurement of nucleolar integrity and p53 intensity (and any other markers of p53-independent NSP activity) would aid in our understanding of the contribution of p53-dependent and independent mechanisms to the overall response of the nucleolus when subjected to stress. On a technical level, we have utilized a cell model that has previously been used in the literature to study the p53-dependent NSP (Bursac et al., 2012; Fumagalli et al., 2009, 2012), which was amenable to high-throughput functional (siRNA) screening and to conducting high-content imaging (nuclear p53 intensity) as a readout. It is possible that some of the candidates we have identified from our screening approach may be specific to the cell line that we have used; for specific candidates identified, we have aimed to validate these results in a different cell line, but we are unable to further comment on the universality of other candidates identified to the overall p53-dependent NSP mechanism, which would require further interrogation outside of the scope of this investigation. Finally, a well-known caveat of siRNA screening is that there may be no phenotype at the given endpoint (as the candidate is not sufficiently depleted, i.e., false negative) or, conversely, off-target effects of the reagents (causing false positives); while we have validated individual candidates using independent candidate-based approaches to ensure target-specific knockdown and level of p53 stabilization, it is possible that some of the factors may not have been identified from our studies. Further unbiased evaluation of all genes associated with specific pathways identified in our study (e.g., ribosome, nucleolus, proteasome, RNA splicing, cell cycle, and Pol II) across a broad range of cell lines would go some way to further validating our observations and to identifying those that are universally required in the NSP.
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Professor Ross Hannan ([email protected]).
Any unique/non-commercially available items generated in this study will be available from the lead contact upon reasonable request.
A549 (human lung adenocarcinoma, CCL-185), HEK-293 (CRL-1573) and MCF10A (CRL-10317) cell lines were obtained from the ATCC. The Flp-In™ T-Rex™ U2OS cells (Al-Hakim et al., 2012) were a generous gift from Prof. Laurence Pelletier (Lunenfeld-Tanenbaum Research Institute, Toronto, Canada). All cell lines were cultured at 37°C with 5% CO2 in complete growth media (outlined below). The cell lines that were used have not been authenticated.
All animal experimentation was performed with approval from the Australian National University Animal Experimentation Ethics Committee (protocol numbers A2015/58, A2018/56). All animals were housed in animal rooms are automatically controlled to have a 12-hour light-dark cycle, and to maintain a constant temperature of approximately 22°C, and animal welfare checked at least once daily. Mouse embryonic fibroblasts were generated from either C57BL/6 Mdm2C305F/C305F mice (a gift from Prof Yanping Zhang, University of North Carolina, Chapel Hill, USA) or their wild-type C57BL/6 littermates after plug mating of homozygous adult mice (females >10 weeks of age) to achieve either wild-type or homozygous mutant progeny. Mouse embryos (unsexed) were isolated from pregnant females at E13.5 and cultured as described below.
Unless otherwise stated, cell culture reagents were purchased from Gibco (ThermoFisher Scientific). A549 cells were cultured in Dulbecco’s Modified Eagle Medium: Nutrient Mixture F-12 (DMEM:F12) with HEPES, supplemented with 10% foetal bovine serum (JRH Biosciences #12003C or Sigma-Aldrich #F9423) and 2 mM GlutaMAX™ or 2 mM L-Glutamine. HEK-293 cells were cultured in DMEM supplemented with 10% FBS (Sigma-Aldrich #F9423) and 2mM L-glutamine. MCF10A cells were cultured in DMEM:F12 containing 5% horse serum (#16050-122), 20 ng/mL epidermal growth factor (EGF, Peprotech), 0.5 mg/mL hydrocortisone (Sigma-Aldrich #H-0888), 100 ng/mL cholera toxin (Sigma-Aldrich #C-8052) and 10 μg/mL insulin (available from pharmacy). The Flp-In™ T-Rex™ U2OS cells were cultured in DMEM supplemented with 10% FBS and 1 μg/mL Blasticidin (Melford). For expression of MDM2 protein, the cDNA for the human full-length MDM2 sequence was cloned into the pcDNA5 vector (ThermoFisher Scientific) to enable expression of the proteins with an N-terminal 2xFLAG-PreScission protease site-His6 (FLAG) tag. The plasmid was then transfected into the cells for stable integration into the host genome and selected using Hygromycin B (Formedium) according to the manufacturers’ instructions.
To generate MEFs, C57BL/6 Mdm2C305F/C305F mice or their wild-type C57BL/6 littermates were plug mated. Mouse embryos were isolated at E13.5 and prepared as similarly described (Durkin et al., 2013), digesting individual embryos in 1 mL of 0.25% trypsin-EDTA (Gibco) containing 100U DNaseI (Sigma-Aldrich #D5025) at 37°C at 10% CO2 for 15 minutes. To stop the digestion, 1 mL of DMEM (Gibco) supplemented with 10% FBS (Sigma-Aldrich #F9423), 1x Antibiotic-Antimycotic, 2 mM GlutaMAX™ and 1x MEM non-essential amino acids (NEAA) was added, and the cells were centrifuged at 300 x g for 5 minutes to pellet the cells. Cell pellets were then resuspended and cultured in fresh DMEM containing 10% FBS, 1x Antibiotic-Antimycotic, 2 mM GlutaMAX™ and 1x MEM NEAA at 37°C.
The α-p53 (human, DO-1, sc-126) antibody was obtained from Santa Cruz Biotechnology. The α-p21 (12D1, #2947S), α-p53 (mouse, 1C12, #2524S) and α-Myc-tag (9B11, #2276) were purchased from Cell Signalling Technologies. The α-RPL5 (ab137617), α-RPL11 (ab79352), α-RPS18 (ab91293), α-RPL21 (ab215724), α-RPS19 (ab57643), α-RPL28 (ab193164) and α-RPL22 (ab77720) antibodies were procured from Abcam. The α-BrDU (B44, #347580) was obtained from BD Biosciences. The α-Actin antibody (C4, #691001) was acquired from MP Biomedicals, while the α-FLAG (F-3165) antibody was purchased from Merck Millipore. The goat α-mouse (#1706516) and α-rabbit (1706515) HRP-conjugated IgG (H+L) antibodies were obtained from BioRad. Alexa-Fluor 594 donkey α-mouse IgG (H+L, A21203) and Alexa-Fluor 488 goat α-mouse IgG (H+L, A11001) were purchased from ThermoFisher Scientific. Fluorescein isothiocyanate (FITC) conjugated sheep α-mouse IgG (Cappel, ICN Biomedical #55516) and horse anti-mouse IgG (FI-2000, Vector Labs) were also purchased.
Actinomycin D (A9415), α-Amanitin (A2263), Bovine Serum Albumin (BSA), 5-bromo-2′-deoxyuridine (BrdU, B5002), Cycloheximide, DAPI (4′,6-diamidino-2-phenylindole), Diethyl pyrocarbonate (DEPC), Dimethyl Sulfoxide (DMSO), Doxorubicin (D1515), 5-Fluorouracil (F6627), Leptomycin B (L2913), MG132 (C2211), Propidium iodide (PI), Sucrose, Tergitol/IGEPAL (NP-40), Triton™ X-100, Tween®20 were purchased from Sigma-Aldrich. Camptothecin (S1288) and Etoposide (S1225) were purchased from Selleck Chemicals and CX-5461 (SYN-3031) was obtained from SYNkinase. Dulbecco’s phosphate-buffered saline (dPBS, pH 7.4, Gibco) and ultrapure water (Invitrogen) were procured from ThermoFisher Scientific. 5X siRNA buffer (B-002000-UB-100, diluted to 1X with ultrapure water), DharmaFECT1 transfection reagent (T-2001) and assay-specific siRNAs (further details provided below) were purchased from Horizon Discovery. Polyethylenimine (PEI, #23966) was purchased from Polysciences. Paraformaldehyde (PFA, diluted in dPBS prior to use) was obtained from Electron Microscopy Sciences (Fisher Scientific). Complete protease inhibitor cocktail and PhoSTOP phosphatase inhibitor were purchased from Roche. All compounds used for treatment of cells, apart from α-Amanitin (H2O), CX-5461 (50 mM NaH2PO4, pH 4.5) and Leptomycin B (70% methanol) were solubilised in DMSO and stored at −20°C in aliquots prior to use.
For candidate-based studies, A549 cells were reverse transfected with up to 50 nM (final concentration) of candidate-specific siRNA (Horizon Discovery, please see Tables S1 and S6 for catalogue numbers for each specific candidate) or non-targeting siRNA (ONTARGETplus (OTP) non-targeting control siRNA, Horizon Discovery, D-001810-10-50, prepared as per manufacturers’ instructions). Briefly, transfection mixes containing DMEM:F12 basal media with DharmaFECT1 (T-2001, final concentration of 1μL/mL) were prepared with siRNA and allowed to complex for 20 minutes. A549 cells (prepared in suspension in DMEM:F12 growth media) were then added on top and the cells incubated at 37°C in 5% CO2. At 24 hours’ post transfection, media was removed and replaced with fresh growth media. For the transfection of U2OS Flp-In cells, siRNAs (final concentration of 50 nM) were prepared by complexing with LipofectamineTM RNAiMAX transfection reagent (in Opti-MEM media, ThermoFisher Scientific) prior to reverse transfection of cells (as similarly described above).
The primary genome-wide siRNA screening was performed using the Dharmacon Human siGENOME SMARTpool siRNA library (Horizon Discovery, G-005005-E2); the secondary screen using individual duplex siRNA was performed using the Human siGENOME siRNA library (set of 4 duplexes, Horizon Discovery, GU-005005-E2). For the custom libraries, human siGENOME SMARTpool siRNAs were cherrypicked for individual genes (catalogue details available in Tables S1 and S6). Transfection mixes containing DMEM:F12 basal media, DharmaFECT1 (T-2001, final concentration of 1 μL/mL; i.e., 0.038 μL in 37.5μL final volume) and siRNA were complexed for 20 minutes in Corning Costar 384 well optical microplates (#3712). We used 40 nM final concentration for the primary screen, 25 nM final concentration for the secondary screen and either 10 nM of non-targeting siRNA (for the “p53 stabilisation” and follow-up co-depletion screens), 10 nM siRPS19 siRNA (for the “modifiers of ribosomal stress” and follow-up co-depletion screens) or 10 nM siHEATR3, siRPL5 or siRPL11 siRNAs (for the follow-up co-depletion screens; details for siRNAs located in Tables S1 and S6). Using a reverse transfection strategy, A549 cells (750 cells/well prepared in DMEM:F12 growth media) were then delivered on top of the transfection mixtures; plates were briefly centrifuged at 500 x g for 1 minute, and then incubated in a LiCONiC microplate incubator at 37°C with 5% CO2 for 24 hours. At 24 hours post-transfection, media was removed and replaced with fresh DMEM:F12 growth media and returned to the incubator. At 72 hours post transfection, media was removed, and cells were fixed with 4% PFA (25 μL/well) for 10 minutes, washed with dPBS (50 μL/well) and permeabilised with of 0.5% TTX-100 in dPBS (25 μL/well) containing 0.5 μg/mL DAPI for a further 10 minutes. Plates were then washed with dPBS twice (50 μL/well). Cells were immunostained with anti-p53 (DO-1) antibody in a 1% BSA in dPBS solution, washed twice with dPBS, followed by the addition of the Alexa-Fluor secondary antibody for signal detection. After the final dPBS wash, stained cells were left in dPBS (50 μL/well), the plates heat-sealed with foil seals (Agilent PlateLoc) and imaged on either a Cellomics ArrayScan VTi (ThermoFisher Scientific) or a Perkin Elmer Opera Phenix high content imaging microscope. Details for image acquisition and quantitative algorithms can be found in Tables S9 and S10.
As both primary screens were combinatorial screens (i.e. co-transfecting two siRNAs simultaneously), we included two sets of controls within the plate; the ‘health’ controls (which were transfected with one siRNA only, to ensure that the cell viability response was consistent within each screen run, and that co-transfection of siRNAs did not impact on cell viability), and the ‘normalising’ controls for each independent screen (two siRNAs co-transfected, which were also used for quality control and normalisation of data). We also included some ‘mock’ (lipid-only) controls which contained the single siRNA which was being co-transfected with the unknowns on the plates. The approach for analysis of each screen is outlined below. We also utilised a similar approach for data normalisation for all subsequent screening experiments. For the primary screens, within the library (as provided by the manufacturer), there were 55 duplicate wells; we opted to remove any duplicate candidates from the library which were screened more than once, where we removed the duplicate(s) that were the least changed when compared to the respective normalisation strategy for the screen (therefore leaving 18,120 unique candidates screened). Using RNAseq data generated from transfection of A549 cells with either siOTP-NT (“p53 stabilisation” screen) or siRPS19 siRNA (“modifiers of ribosomal stress” screen) for 72 hours, we removed ‘non-expressed’ candidates from the screening datasets (based on an RPKM value cut-off of <0.05). All subsequent analyses were performed on the ‘expressed’ candidates (13,855 and 14,577) in the “p53 stabilisation” and “modifiers of ribosomal stress” screens, respectively.
Each plate contained multiple wells of three different negative control conditions (siOTP-NT, siTP53 and mock transfected) and one positive control (siRPS19). For the normalisation controls, to balance the absolute amount of siRNA and to replicate precise screening conditions, each control was co-transfected with siOTP-NT (i.e., siOTP-NT+siOTP-NT, siTP53+siOTP-NT, siRPS19+siOTP-NT etc). For the health controls, we normalised to the average siOTP-NT value (presented as a fold-change, FC); for the normalisation controls and library candidates, we normalised to the average of the siOTP-NT+siOTP-NT wells (data presented as FC) – this approach was used for both p53 intensity and cell number outputs and used for evaluating quality control metrics. Our QC metrics also focused on specific control combinations – siOTP-NT (negative control which had some baseline p53 expression) and siRPS19 (which enhanced nuclear p53 stabilisation). We calculated the average health (siOTP-NT only) and normalisation (siOTP-NT+siOTP-NT) controls to be 1.02 ± 0.19 (mean % co-efficient of variation for each library plate screened; %CV = 18.49) and 1.02 ± 0.19 (mean %CV = 18.49), respectively. The positive controls (siRPS19 and siRPS19+siOTP-NT) scored 5.86 ± 0.86 (mean %CV = 14.61) and 6.17 ± 0.85 (mean %CV= 13.68), respectively. The Z′-factor calculated for siOTP-NT+siOTP-NT vs siRPS19+siOTP-NT for each plate (approach as similarly described (Birmingham et al., 2009)) demonstrated a very robust result of 0.4 ± 0.16. All data was normalised to the average siOTP-NT+siOTP-NT value for each plate, ensuring we could directly use our experimentally determined 2-fold increase in p53 as the cut-off (Figures S1A–S1D) for further analysis. For graphical presentation of the data, FC data was converted to Log2 (p53 FC value).
We used a similar approach described above for the health (single-siRNA) controls and obtained p53 intensity FC values (normalised to siOTP-NT) of 1.02 ± 0.20 (mean %CV across entire screen = 19.39%) for siOTP-NT and 6.14 ± 1.01 (mean %CV = 16.43%) for siRPS19 overall. For this screen, the control combinations for QC were different, because depletion of RPS19 leads to an extremely robust and strong increase in p53 stabilisation. Therefore, the ‘normalisation’ controls for this screen were the siOTP-NT+siRPS19 (negative control) and the siTP53+siRPS19 (positive control). The data was normalised to the average negative control value (siOTP-NT+siRPS19; 1.01 ± 0.17, mean %CV = 16.63%). The positive control (siTP53+ siRPS19) averaged 0.30 ± 0.06 (mean %CV = 19.47%) across the entire screen. The overall Z′ factor was weaker than the ‘p53 stabilisation screen’ but still significant on average (0.04 ± 0.19). To identify candidates from this screen, we applied a cut-off of 3SD above the average positive control FC readout (siTP53+siRPS19) (calculated to be 0.48; relaxed to 0.5). Based on these criteria, we analysed 64 candidates using the Dharmacon siGENOME SMARTpool individual duplexes (25 nM/duplex). We observed in some cases that the duplexes demonstrated a weaker suppression of p53 induced by RPS19 depletion compared to the SMARTpool screen (likely because multiple siRNAs targeting the same mRNA transcript more efficiently deplete the target), therefore, we relaxed the hit cut-off criteria to a p53 FC value of ≤0.65. We proceeded to further validate candidates that had 2 or more duplexes fit these criteria.
The levels of p53 protein were quantitatively analysed using the Alphascreen SureFire Total p53 Assay Kit (TGR Biosciences/Perkin Elmer #TGRT53). Briefly, A549 cells (40,000 cells/well) were reverse transfected with 40 nM siRNA (final reaction volume 0.5 mL) in Nunc 24-well cell culture plates for 72 hours, refreshing media at 24 hours’ post transfection. Just prior to harvest, to normalise p53 signal, plates were imaged on the IncuCyte S3 Live-Cell Analysis Imaging System (Essen Biosciences) using the scan-on-demand feature (2015A software release), calculating the Phase Object Confluence (Percent) for each well. Media was then carefully removed, and the cell monolayers lysed with 30 μL 1x Alphascreen Lysis Buffer for 2-3 minutes on an orbital rocker, prior to transfer of lysates to −20°C. Lysates (4 μL/well in Perkin Elmer 384 well proxiplates, #6008280) were then assayed in as per the manufacturers’ instructions. Plates were then read using the Perkin Elmer EnSpire Multimode Plate Reader instrument using the instrument pre-defined Alphascreen settings and the data normalised to the cell confluence readings, then normalised to the non-targeting siRNA treated wells.
Either A549 cells (0.9 × 105/well, transfected as described above) or MEFs (∼3 × 105/well) were plated in Nunc 6-well tissue culture dishes and incubated for 24 hours. For determination of candidate-specific knockdown (where appropriate), unless otherwise described, cells were harvested at 24 hours’ post transfection/seeding; otherwise, the transfection/culture media was removed and replenished with either fresh media, or pharmacological inhibitors at the required dose (prepared in growth media) and incubated for the desired period (up to 72 hours). For cells which were UV treated, subjected to heat-shock (45°C) or irradiated (IR) with gamma irradiation, cells were plated down for 69 hours, treated with 5 mJ/cm2 using a UV Crosslinker (UV Stratalinker 1800), heated at 45°C for 30 minutes, or exposed to 10 Gy (Rad Source RS 2000), returned immediately to a 37°C incubator with 5% CO2 and incubated for 3 hours prior to harvested for immunoblotting. To assess p53 stabilisation, A549 cells depleted of RPS19 for 24 hours were treated with 1 μg/mL CHX for 10 minutes, then protein harvested for western blotting as previously described.
Cells were pulse labelled with 10 μM BrdU and incubated for 30 minutes at 37°C 5% CO2. Media was collected, the cell monolayer washed with dPBS then the adherent cells trypsinised for 5 minutes at room temperature with 0.25% Trypsin-EDTA. Trypsinised cells were then neutralised (1:1) with growth media. The media, washes and trypsinised cells were pooled, and the cells centrifuged at 1200 rpm for 5 minutes to pellet the cells. Cells were then fixed with ice-cold 80% ethanol. Prior to FACS analysis, cells were permeabilised (1M HCl, 0.5% TTX-100) then incubated with α-BrDU antibody, followed by FITC/Alexa-Fluor-488 conjugated α-mouse IgG secondary antibody. Cells were then stained with PI (0.01 mg/mL) and analysed on either a BD FACS CantoII or LSR instrument. All FACS data was analysed using FlowLogic Software (Inivai Technologies).
FLAG-tagged human RPL5 and RPL11 (FLAG-RPL5, FLAG-RPL11) constructs have been previously described (Sloan et al., 2013). To generate the human myc-tagged (MT) human HEATR3 construct, the MGC clone containing the full-length human HEATR3 cDNA (MHS6278-202832721) was purchased from GE Life Sciences as a template for PCR cloning. The pcDNA3.1 vector (containing a puromycin selection cassette) was a kind gift from A/Prof Armadeep Dhillon (LaTrobe University, Bundoora, Australia). Primers for cloning were purchased from Sigma-Aldrich. The 5’ (forward) primer (5′-GAATTCGGATCCGCCACCATGGGGGAACAAAAACTCATCTCAGAAGAGGATCTGGGCAAGAGCCGGACGAAG-3′) was engineered to include a 5′ BamHI restriction site, immediately proceeded by a nucleotide sequence that, when translated in-frame, generates a methionine residue and the myc-tag (M-EQKLISEEDL), followed by the endogenous human HEATR3 sequence (residues 2-680). The reverse primer (5′-GAATTCCTCGAGTTAAGAAGTCAGTCTTTTCTCAACAGTTTC-3′) contained a XhoI restriction site immediately after the stop codon. The cDNA template was incubated with the KOD Hot Start Mastermix (Novagen) and 10 μM of forward and reverse primers and amplified in a BioRad Thermal Cycler using the following conditions (95°C for 2 minutes, three cycles at 95°C for 20 seconds, annealing at 62°C for 10 seconds and extension at 70°C for 45 seconds, followed by 33 cycles of denaturation at 95°C for 20 seconds, annealing at 67°C for 10 seconds and extension at 70°C for 45 seconds). The PCR product (2106 bp) was purified using the Nucleospin Gel and PCR Clean-up kit (Macherey Nagel) and along with the pcDNA 3.1 vector, digested at 37°C with BamHI and XhoI restriction enzymes as per the manufacturers’ instructions (Promega) for 1.5 hours. Digests were then incubated at 65°C for 10 minutes to denature the restriction enzymes, before electrophoresing on 1.2% Tris-Acetate-EDTA (TAE) agarose gels containing Sybr-Safe (ThermoFisher Scientific) at 100V for ∼1.5 hours. The digested insert and vector were then extracted from the gel and cleaned up using the Macherey-Nagel Gel and PCR Clean-Up Kit (as per manufacturers’ instructions). The concentration of the DNA was determined using a Nanodrop 2000 instrument (ThermoFisher Scientific), and a ligation reaction prepared as per manufacturers’ instructions using the Promega T4 DNA Ligase Kit at a 1:3 (vector: insert) ratio. The reaction was then incubated at 65°C for 10 minutes to denature the enzyme and cooled on ice for 2 minutes. The ligation reaction was added to Bioline chemically competent bacteria (BIO-85027), incubated on ice for 30 minutes, heated at 42°C for 90 seconds and then placed on ice for 2 minutes. Cells were recovered in 1 mL of Super Optimal broth with Catabolite repression (SOC) media (Bioline) for 1 hour at 37°C shaking at 200 rpm prior to plating (under aseptic conditions) onto L-Broth (LB) Agar plates containing 100 μg/mL ampicillin and incubated for 16 hours at 37°C. Individual colonies were picked and inoculated into 5 mL LB containing 100 μg/mL ampicillin and incubated, shaking at 200 rpm, for 16 hours at 37°C. From 2 mL of each culture, plasmid DNA was extracted using the Nucleospin Plasmid Kit (Macherey-Nagel) as per manufacturers’ instructions. The MT-HEATR3 constructs were sequenced verified by Sanger sequencing (Micromon Genomics, Monash University, Australia) prior to use.
Unless otherwise described, total RNA was harvested and isolated using the NucleoSpin RNA extraction kit (Macherey-Nagel), Qiagen RNeasy Mini kit, or the Bioline Isolate II RNA mini kit, according to the manufacturer’s instructions. Total RNA was quantified using the Nanodrop 2000C Spectrophotometer (Thermo Scientific) prior to storage at −80°C. To measure total 28 and 18S rRNA in samples, or to perform sample quality control prior to RNAseq experiments, total RNA was subjected to analysis using the Agilent 2100 Bioanalyser and the Agilent RNA Nano 6000 kit as per the manufacturers’ instructions. cDNA was prepared using the Invitrogen Superscript III Kit (ThermoFisher Scientific) with random primers (Promega) as previously described (George et al., 2013).
Primer design and quantitative real-time PCR (qPCR) analysis was carried out as similarly described (George et al., 2013). Where appropriate, gene expression was normalised to housekeeping gene expression (GAPD or B2M) and fold change determined using the relative quantitation (2-(ΔΔCt)) method (Livak and Schmittgen, 2001). To measure TP53 mRNA from polysome fractions, we calculated the percentage of RNA in each fraction utilising a method as previously described (Han et al., 2022). A list of primer sequences used in this publication is in Table S8.
Total RNA for sequencing was isolated from cells using the RNeasy Mini Kit (Qiagen) according to the manufacturers’ instructions, and concentration and quality determined using the Agilent 2100 Bioanalyser with the Agilent RNA Nano 6000 kit as previously described. Sequencing libraries were prepared using the TruSeq RNA library preparation kit (Illumina) and sequenced on an Illumina HiSeq2500 (6 samples/lane). The generated 50bp paired-end reads were aligned to the genome using TopHat (Trapnell et al., 2009) v2.0.8b with default parameters and the reads counted using HTSeq (Anders et al., 2015). The differential expression was calculated utilising the DESeq package (Anders and Huber, 2010) in R (version 3.0.2). Absolute gene expression was defined by determining reads per kilobase per million (RPKM) as previously described (Mortazavi et al., 2008).
Unless otherwise described, cell monolayers were washed with dPBS then lysed in Western Solubilisation Buffer (20 mM HEPES, 0.5 mM EDTA, and 2% SDS). Protein concentration was determined using BioRad DC Assay as per the manufacturers’ instructions. Proteins were separated by SDS-polyacrylamide gel electrophoresis, transferred to PVDF membrane (Immobilon-P, Merck Millipore), blocked with 5% skim milk (prepared with Tris-buffered saline containing 0.1% Tween-20; TBST) at room temperature, and incubated with primary antibody either at room temperature (1-2 hours) or overnight at 4ᵒC. Membranes were washed with TBST, incubated with secondary antibodies for 1 hour, then the protein bands visualised using Clarity Western ECL (BioRad) or Western Lightning ECL (Perkin Elmer) and images either acquired digitally using the Bio-Rad Gel Doc System (with Image Lab Touch Software) or onto Amersham Hyperfilm (Cytivia/GE Healthcare).
Cells were seeded and/or transfected in CC2-coated 4-well chamber slides (Thermo Fisher Scientific). Cells were fixed at the desired timepoint with 4% PFA, rinsed with dPBS and permeabilised with 0.5% TTX-100 prepared in dPBS containing 0.5 μg/mL DAPI. Slides were rinsed twice with dPBS and incubated with the primary antibody diluted in 1% BSA prepared in dPBS for 1 hour at room temperature, followed by washing with dPBS twice then incubating with the appropriate AlexaFluor-conjugated secondary antibody for 1 hour. After rinsing with dPBS twice, the chamber was removed and a coverslip with Vectashield mounting medium (Vector Laboratories) was placed over the cells and sealed with clear nail polish. Images were captured using an Olympus BX-51 (widefield, 40X objective) using SPOT Advanced software (v4.6, Diagnostic Instruments) and analysed/pseudocoloured images using NIH ImageJ (v2.1.4.6) (Rueden et al., 2017; Schneider et al., 2012) and Adobe Photoshop (CS6).
Information about where specific ribosomes integrate into the eukaryotic ribosome was obtained from a paper published by de la Cruz and colleagues (de la Cruz et al., 2015). We rescreened the majority of RP genes in a boutique siRNA screen (as similarly described above for the primary screen analysis, but encompassed several RPs which were not screened in the primary screen; details in Table S5), and further stratified the times at which the specific RP was integrated into the ribosome into 6 groups: Early Nucleolar and Nucleolar (Early Nucleolar/Nucleolar), Medium Nucleolar, Late Nucleolar (Medium/Late Nucleolar), Early Nucleolar – Late Positioning, Late Nucleolar – Cytoplasmic, Cytoplasmic and Unknown (Table S5). The p53 intensity data for each RP was then binned according to its assembly group. For mapping of the p53 values for RPs onto the ribosome, we imported the existing 80S human ribosome structure published by Khatter and colleagues (PDB reference 4UGO)#(Khatter et al., 2015) into the Pymol Molecular Graphics System (Schrödinger, LLC) and utilised Pymol colours in the red and grey spectrums (ranging from the highest p53 value “firebrick” to the lowest value “grey90”) to manually colour each RP (using data in Table S5) within the ribosome structure. Note that RPLP0, RPLP1, RPLP2 and RPLP12 are missing from the structure (and are therefore not included), and RPSA and RACK1 were not rescreened for this analysis.
U2OS cells expressing FLAG-MDM2 were subjected to siRNA-mediated knockdown of HEATR3 for 48 hours, then treated with 1mg/mL tetracycline (Duchefa Biochemie) for 16 hours to induce MDM2 protein expression. Cells were then harvested by sonication in gradient buffer E (20 mM HEPES pH 8.0, 150 mM KCl, 0.5 mM EDTA, 0.1 mM DTT, 5% glycerol) and loaded onto a 10-40% glycerol gradient containing gradient buffer (20 mM HEPES pH 8.0, 150 mM KCl, 1.5 mM MgCl2, 1 mM DTT, 0.2% Triton X-100). The gradients were centrifuged in a swTi60 rotor (Beckman L7-80) at 4°C at 52,000 rpm for 1.5 hours. The gradient was then manually fractionated into 200 μL fractions. The non-ribosomal (free) fractions at the top of the gradient (fractions 1-6) were then pooled and immunoprecipitated with either control IgG Sepharose beads or anti-FLAG antibody beads (Sigma-Aldrich) using gradient buffer containing 10% glycerol. Co-purified RNA was extracted using phenol/chloroform/IAA, ethanol precipitated, separated on an 8% acrylamide/7M urea gel, transferred to a Hybond-N membrane (GE-Healthcare) and then analysed by northern blotting using a radiolabelled probe (5′-CCGAGATCAGACGAGATCGGGCGCGTTCAGGGTGGTATGG-3′) hybridizing to the 5S rRNA (as previously described (Sloan et al., 2013). Densitometry data (ImageJ) was analysed by comparing the ratio of signal in the FLAG-IP compared to the input sample, and then expressed as a percentage of non-targeting transfected cells.
HEK293 cells (200,000/well) were seeded into Nunc 6-well tissue culture plates and incubated overnight at 37°C in 5% CO2. The following day, a total of 0.5 μg of plasmid DNA per well was complexed with PEI (1:4.5 ratio of plasmid DNA:PEI) in DMEM basal media for 20 minutes, prior to addition to the cells (added dropwise to the media). Cells were then incubated at 37°C in 5% CO2 for 48 hours. Transfected cells were washed once with dPBS, and then harvested in CoIP buffer (100 mM Tris pH 8.0, 100 mM NaCl, 1% NP40, Complete protease inhibitor cocktail and PhoSTOP), swirling gently on an orbital rocker for 1 hour at 4°C. Cell lysates were then collected and centrifuged at 14,000 rpm for 10 minutes at 4°C; the supernatant was then collected, and the concentration determined using the BioRad DC assay as per manufacturers’ instruction. Each CoIP sample (300 ng of protein) was precleared using 10 μL of Novex Protein G Dynabeads (Life Technologies, #10004D) and rotated at 4°C for 1 hour. The pre-cleared lysate was then transferred to eppendorf tubes containing 3 μg anti-FLAG antibody diluted in 1x CoIP buffer and incubated overnight, rotating at 4°C. Dynabeads were then washed four times with 1 mL of ice-cold CoIP buffer. The sample was eluted off the beads by boiling at 95°C for 5 minutes in 25 μL of 2x SDS-PAGE loading buffer (130 mM Tris-HCl, pH 6.8, 0.67% (w/v) SDS, 16% (v/v) glycerol, 0.083% (w/v) bromophenol blue and 19.3 mM β-mercaptoethanol), electrophoresed on 4-20% Tris-Glycine gradient gels (Invitrogen) and immunoblotted as described above.
A549 cells (transfected with target-specific siRNAs as described above for 72 hours in 10 cm2 dishes, seeded at 3.2 × 105 cells/plate) were treated with 100 μg/mL cycloheximide (CHX) for 5 minutes then harvested in hypotonic lysis buffer as similarly described (Chan et al., 2011). Lysates from equal cell number (8 × 106) were loaded onto high salt (250 mM NaCl, 3.1–30.1%) or low salt (80 mM NaCl, 10-40% (w/v)) sucrose gradients generated using a BioComp Gradient Master 108 and separated by centrifugation (SW41 rotor at 40,000 rpm for 4 hours or 36,000 rpm for 2.15 hours, respectively) using a Beckmann Coulter Optima XE-100 Ultracentrifuge. Samples were fractionated (1 mL fractions) using a Teledyne ISCO Foxy R1 instrument. Absorbance at 260 nm was determined using a Brandel UA-6 UV/Vis detector, and measurement of the area under the peaks in each trace conducted. For conducting analysis of transcripts being translated, RNA was isolated from individual fractions using the Qiagen RNeasy kit as per manufacturers’ instructions. cDNA was then synthesised from samples and analysed using qPCR as described in the relevant sections above.
A549 cells were reverse transfected with siRNAs (as per siRNA transfection procedure above) for up to 72 hours. At approximately 2 hours prior to pulsing cells with 32P orthophosphate (Perkin Elmer, #NEX053H025MC), cells were phosphate starved in phosphate-free DMEM (Thermo Fisher #11971025) containing 10% dialysed FBS (ThermoFisher #26400044) and 2mM GlutaMAXTM for 2 hours at 37°C with 5% CO2. Cells were then pulsed with between 3-5 mCi of 32P orthophosphate per sample, then incubated for between 30-40 minutes at 37°C with 5% CO2. Cells were then gently rinsed with pre-warmed PBS and replenished with A549 growth media and incubated for a further 30 minutes at 37°C with 5% CO2. Total RNA was then harvested, purified, and quantified using the Nanodrop as describedabove. RNA samples to be electrophoresed were heated to 65°C for 15 minutes, quenched on ice, and ∼1mg/mL of ethidium bromide added per sample. Samples were then subjected to MOPS/formaldehyde (1.2%) gel electrophoresis for 16-24 hours at 45V. Gels were imaged using theBio-Rad Gel Doc System, then dried under vacuum for ∼4 hours. Dried gels were then exposed onto phosphoscreen for between 12 hours – 5 days and imaged on the Typhoon FLA9000 Phosphorimager (GE Healthcare Life Sciences). Densitometry analysis of bands corresponding to the 32P labelled RNA was then conducted using ImageJ.
An initial alignment of human HEATR3 with Cheatomium thermophilum Syo1 (ctSyo1) was extracted from an HHPRED search (Soding, 2005; Zimmermann et al., 2018). The ctSyo1 crystal structures PDB: 4GMO (free) and PDB: 5AFF (in complex with RPL5/11) scored E-values <e−75 over the entire protein sequence despite low sequence identity. To refine the assignment, a consensus secondary structure prediction of HEATR3 was created using PSIPRED4 (Buchan and Jones, 2019) and Quick2D (Soding, 2005). Subsequently, ctSyo1 secondary structure features (helices of over three residues in length) were extracted from the crystal structures and manually aligned with HEATR3 predictions while maintaining primary sequence similarities using the program ALINE (Bond and Schuttelkopf, 2009). Additionally, high-confidence prediction of disordered regions (Jones and Cozzetto, 2015) with a DISOPRED 3 score higher than 0.5 (over a ten residue stretch) were used to assist in this process. The manual alignment based on secondary structure-predictions was used to generate a 3D model of HEATR3 with the MODELLER 9.22 suite (Webb and Sali, 2016) using PDB 4GMO and 5AFF as templates. The obtained models only diverged in the contraction state between N- and C-terminal regions of the chain, as observed between free and RPL5/11 bound ctSyo1. Positions of RPL11 and the conserved RPL5 N-terminus were modelled based on their interaction with ctSyo1 (PDB 5AFF) (Kressler et al., 2012). Figures were prepared using the Pymol Molecular Graphics System (Version 2.0, Schrödinger, LLC, https://pymol.org/2/).
Biological networks were identified from the screening dataset using STRING (versions 9-11) software (http://string-db.org) (Franceschini et al., 2013; Szklarczyk et al., 2019), a publicly available online database of physical and functional protein interactions, using the ‘high confidence interactions’ setting and the KEGG network enrichment analysis feature. Given its large size, in order to organise the network relative to pathways and functional annotation, data obtained from STRING from the “p53 stabilisation” primary screen (Table S2) with an RPKM value of ≥0.05 and fold change (FC) p53 value of ≥2.0 was plotted using Cytoscape (www.cytoscape.org/), an open source platform for complex network analysis (Shannon et al., 2003) after applying a “force directed layout”.
A gene ontology (GO) enrichment analysis was performed for the genes identified in the “p53 stabilisation screen” with an RPKM ≥0.05 and FC (p53) ≥ 2.0 (Table S1) as the query genes, versus all genes from the dataset (as the gene universe) using the ClusterProfiler package (Yu et al., 2012) in R (R Core Team (2018) R: a language and environment for statistical computing). We considered three main ontologies: “biological process” (BP), “cellular component” (CC) and “molecular function” (MF); the enrichment p-values are corrected for multiple hypothesis testing using the Benjamini-Hochberg procedure within the package, and all the GO terms that were further considered were statistically significantly enriched at the α = 0.01 level. We then further simplified the GO terms per ontology by removing semantic redundancies of terms using the default method (proposed by Wang and colleagues (Wang et al., 2007)) which uses the topology of the GO graph structure to compute semantic similarity. We then visualised the top 20 GO terms per ontology across the BP, CC and MF groups (BP is demonstrated in Figure 1D of this manuscript).
A cell localisation enrichment analysis was performed for the candidates identified in the “p53 stabilisation screen” with an RPKM ≥0.05 and FC (p53) ≥ 2.0 (Table S4) as the query genes, versus all genes from the dataset (gene universe) using the LOCATE subcellular localisation database (Sprenger et al., 2008), which lists the subcellular localisation of 64,637 human proteins (https://web.archive.org/web/20171231015119/) http://locate.imb.uq.edu.au/. The full human database was imported into R software, then converted and parsed to store as an R-readable object. From there, we extracted the subcellular localisation information for all query genes and genes from the gene universe and performed an enrichment analysis of the number of query genes in every subcellular localisation category. The results were then visualised using a dotplot, where the size of the dot indicates the number query genes found in the localisation category. The log10 odds-ratio (OR) reflects the amount of enrichment/under representation, i.e., log10 OR < 0 indicates under-representation, and log10 OR > 0 indicates over-representation of query genes in the corresponding category. The coloured bars represent the 95% confidence intervals.
Unless otherwise described, statistical tests were performed using GraphPad Prism Software (https://www.graphpad.com) with the relevant statistical test details (including any post-hoc analyses), sample size and/or replicate number described for each analysis at the appropriate location in the text (either within the main text and/or figure and supplementary figure legends). |
|
PMC9647044 | Lisa E.L. Romano,Wen Yih Aw,Kathryn M. Hixson,Tatiana V. Novoselova,Tammy M. Havener,Stefanie Howell,Bonnie Taylor-Blake,Charlotte L. Hall,Lei Xing,Josh Beri,Suran Nethisinghe,Laura Perna,Abubakar Hatimy,Ginevra Chioccioli Altadonna,Lee M. Graves,Laura E. Herring,Anthony J. Hickey,Konstantinos Thalassinos,J. Paul Chapple,Justin M. Wolter | Multi-omic profiling reveals the ataxia protein sacsin is required for integrin trafficking and synaptic organization | 01-11-2022 | ARSACS,sacsin,proteomics,cell surface,microtubules,focal adhesions,synaptic adhesion proteins,integrins,Purkinje neurons,synapse | Summary Autosomal recessive spastic ataxia of Charlevoix-Saguenay (ARSACS) is a childhood-onset cerebellar ataxia caused by mutations in SACS, which encodes the protein sacsin. Cellular ARSACS phenotypes include mitochondrial dysfunction, intermediate filament disorganization, and progressive death of cerebellar Purkinje neurons. It is unclear why the loss of sacsin causes these deficits or why they manifest as cerebellar ataxia. Here, we perform multi-omic profiling in sacsin knockout (KO) cells and identify alterations in microtubule dynamics and mislocalization of focal adhesion (FA) proteins, including multiple integrins. Deficits in FA structure, signaling, and function can be rescued by targeting PTEN, a negative regulator of FA signaling. ARSACS mice possess mislocalization of ITGA1 in Purkinje neurons and synaptic disorganization in the deep cerebellar nucleus (DCN). The sacsin interactome reveals that sacsin regulates interactions between cytoskeletal and synaptic adhesion proteins. Our findings suggest that disrupted trafficking of synaptic adhesion proteins is a causal molecular deficit in ARSACS. | Multi-omic profiling reveals the ataxia protein sacsin is required for integrin trafficking and synaptic organization
Autosomal recessive spastic ataxia of Charlevoix-Saguenay (ARSACS) is a childhood-onset cerebellar ataxia caused by mutations in SACS, which encodes the protein sacsin. Cellular ARSACS phenotypes include mitochondrial dysfunction, intermediate filament disorganization, and progressive death of cerebellar Purkinje neurons. It is unclear why the loss of sacsin causes these deficits or why they manifest as cerebellar ataxia. Here, we perform multi-omic profiling in sacsin knockout (KO) cells and identify alterations in microtubule dynamics and mislocalization of focal adhesion (FA) proteins, including multiple integrins. Deficits in FA structure, signaling, and function can be rescued by targeting PTEN, a negative regulator of FA signaling. ARSACS mice possess mislocalization of ITGA1 in Purkinje neurons and synaptic disorganization in the deep cerebellar nucleus (DCN). The sacsin interactome reveals that sacsin regulates interactions between cytoskeletal and synaptic adhesion proteins. Our findings suggest that disrupted trafficking of synaptic adhesion proteins is a causal molecular deficit in ARSACS.
Autosomal recessive spastic ataxia of Charlevoix-Saguenay (ARSACS) is a childhood-onset neurological disease characterized by pyramidal spasticity, cerebellar ataxia, and Purkinje cell loss, which is thought to have both neurodegenerative and neurodevelopmental components (Vermeer et al., 1993). ARSACS was initially believed to be restricted to the Charlevoix-Saguenay region of Quebec, Canada, due to a founder effect mutation (Bouchard et al., 1978). However, since the discovery of the causal gene, more than 170 distinct mutations in SACS have been identified worldwide, and ARSACS is now estimated to be the second most common autosomal recessive cerebellar ataxia (Engert et al., 2000; Synofzik et al., 2013). Sacsin/DNAJC29 expression is ubiquitous but is especially high in large neurons in brain regions associated with motor systems, including layer-V pyramidal neurons in the motor cortex and cerebellar Purkinje cells (Saunders et al., 2018). Sacsin is a large 520 kDa modular protein with domains that implicate it in molecular chaperone and protein quality control systems (Anderson et al., 2010; Parfitt et al., 2009). These include an N-terminal ubiquitin-like domain, regions of homology to the ATPase domain of Hsp90, and a functional J-protein domain, suggesting that sacsin has the ability to modulate Hsp70 chaperone activity. However, the large size of sacsin has hampered biochemical and structural investigations into its function. Patient-derived fibroblasts and sacsin knockout (KO) cell models demonstrate reorganization of the vimentin intermediate filament cytoskeleton, altered mitochondrial network dynamics and trafficking, decreased mitochondrial respiration, and increased mitochondrial stress (Bradshaw et al., 2016; Duncan et al., 2017; Gentil et al., 2019; Girard et al., 2012; Lariviere et al., 2015). Aptamer-based proteomics in sacsin KO SH-SY5Y neuroblastoma cells also found altered expression of proteins involved in synaptogenesis and cell engulfment (Morani et al., 2020). Sacs(−/−) mice recapitulate the motor deficits and cerebellar atrophy observed in ARSACS patients, and undergo progressive age-dependent loss of cerebellar Purkinje neurons, abnormal bundling of non-phosphorylated neurofilament (Lariviere et al., 2015, 2019), and changes to the structure of Purkinje neuron synapses in the deep cerebellar nucleus (DCN) (Ady et al., 2018). While these cellular phenotypes may affect neuron function and survival, their precise relationship to neurodegeneration in ARSACS is unclear. For example, diverse neurodegenerative diseases exhibit altered mitochondrial dynamics and intermediate filament phenotypes (Didonna and Opal, 2019; Stanga et al., 2020), although whether these phenotypes are causal or merely components of a conserved neurodegenerative cascade is an important unanswered question (Gan et al., 2018). Here, we take a multi-omic approach to determine how the loss of sacsin causes these phenotypes and why this disease manifests as a cerebellar ataxia. Our data suggest that altered trafficking of synaptic adhesion proteins is a causal molecular deficit in ARSACS.
To understand the molecular deficiencies in ARSACS, we generated a sacsin KO human SH-SY5Y cell line (Figure S1A), which is widely used to model neurodegenerative diseases (Xicoy et al., 2017). Consistent with ARSACS patient fibroblasts (Duncan et al., 2017) and Sacs(−/−) mice (Lariviere et al., 2015), KO cells had abnormal bundling and asymmetric partitioning of multiple intermediate filaments, including vimentin (Figures 1A and S1B), neurofilament heavy, and peripherin (Figures S1C–S1E). As phosphorylation is a key post-translational modification controlling intermediate filament assembly and disassembly (Snider and Omary, 2014), we performed quantitative proteomic and phosphoproteomic profiling of sacsin KO cells (Table S1). We identified decreased abundance of several proteins previously described in ARSACS patient fibroblasts, including vimentin, the mitochondrial protein ATP5J, and the autophagy-regulated scaffold SQSMT1/p62 (Duncan et al., 2017) (Figure 1B and Table S1). Among the overabundant proteins were the tau-tubulin kinase 1 (TTBK1) and microtubule-associated protein tau (MAPT) (Figures 1B and S1F–S1I), which was hyperphosphorylated at several sites (Figures S1J–S1L). To assess the functional significance of each phosphosite, we analyzed our data in light of a recent machine-learning approach that estimated the effects of individual phosphosites on organism fitness (Ochoa et al., 2020). This analysis identified several highly functional hypophosphorylated residues in vimentin and the nuclear lamina intermediate filaments LMNA/LMNB2 (Figure 1C), which is intriguing considering that ARSACS neurons have altered nuclear shape and positioning (Duncan et al., 2017). Other hypophosphorylated proteins included the focal adhesion (FA) protein zyxin (ZYX) and ataxin 2-like protein (ATXN2L). In addition to tau, several other microtubule-regulating proteins were hyperphosphorylated, including the primary cilia protein ARL3 (Zhou et al., 2006), and the scaffold stathmin (STMN1), which promotes microtubule assembly in a pS16-dependent fashion (Di Paolo et al., 1997). When analyzing changes in phosphorylation corrected for changes in total protein levels, the most hypophosphorylated proteins were RPS6, NLM1, and ATXN2L, which have been implicated in neuronal autophagy and likely reflect increased autophagy in sacsin KO cells (Figure S1M) (Duncan et al., 2017; Key et al., 2020; Klionsky et al., 2021; Tang et al., 2021). The most hyperphosphorylated residues were again in microtubule-related proteins, such as HN1/JPT1 and ARL3. In all, these results suggest that altered phosphorylation may be a contributing factor to cellular ARSACS phenotypes. Kinases are attractive drug targets (Krahn et al., 2020) but are typically lowly expressed and difficult to detect with standard proteomics. Therefore, we enriched for kinases using multiplexed kinase inhibitor beads and performed quantitative mass spectrometry (Cooper et al., 2013). The kinome was broadly altered in sacsin KO cells (Figures S1N and S1O; Table S1). Interestingly, specific families were generally misexpressed in similar directions. For example, the tyrosine kinase family (TK) members were generally downregulated, while CMGC family members were generally upregulated (Figure 1D). Strikingly, we identified ten overexpressed kinases which directly phosphorylate tau at residues that were hyperphosphorylated in sacsin KO cells (Figures 1E and S1N). The most overabundant kinase, BRSK2, and additional CAMK family members MARK1/2/3, all phosphorylate Ser262 in the microtubule-binding domain of tau (Ando et al., 2016; Kishi et al., 2005) (Figures 1E–1H). Phosphorylation of tau Thr231 by DYRK1A is also associated with the detachment of tau from microtubules (Coutadeur et al., 2015; Sengupta et al., 1998). In pathological settings, tau overabundance and hyperphosphorylation can cause the aggregation of insoluble tau and the formation of neurofibrillary tangles. However, we did not find evidence of increased tau aggregation in either undifferentiated or neuronally differentiated sacsin KO cells (Figure S1P). Yet, independent of aggregation, tau phosphorylation can affect microtubule stability, interfere with motor protein function, and disrupt axonal trafficking (Dixit et al., 2008; Ikezu et al., 2020; Stoothoff and Johnson, 2005). Combined with the altered phosphorylation of other microtubule-related proteins, these data suggest that microtubule structure or function may be altered in sacsin KO cells.
We next sought to determine whether microtubule structure and function are affected in sacsin KO cells. We found that cage-like vimentin bundles form around γ-tubulin, a marker of the microtubule organizing center (MTOC), which is a central hub for microtubule nucleation and cargo transport (Martin and Akhmanova, 2018) (Figures 2A and S2A). Acetylated α-tubulin, a microtubule-stabilizing post-translational modification, was increased in sacsin KO cells without affecting total α-tubulin distribution or level (Figures 2B–2D). To assess microtubule dynamics, we treated cells with the microtubule destabilizer nocodazole and found enhanced microtubule polymerization following nocodazole washout (Figures 2E and 2F). Sacsin KO cells also demonstrated increased microtubule polymerization and disordered movements as assessed by live cell imaging of the microtubule plus-end binding protein EB1:GFP (Figures 2G and 2H; Videos S1 and S2). Video S1. Live cell TIRF microscopy in EB1:GFP-labeled SH-SY5Y cells (WT left, KO right) Video S2. Live cell spinning-disk confocal microscopy in EB1:GFP-labeled SH-SY5Y cells (WT left, KO right) Mitochondrial trafficking in neurons is dependent on microtubules (Melkov and Abdu, 2018), and tau overexpression and hyperphosphorylation can cause decreased mitochondrial trafficking (Ando et al., 2016; Lopes et al., 2017; Reddy, 2011), buildup of mitochondria around the MTOC (Ebneth et al., 1998), and DRP1 mislocalization and reduced mitochondrial fission (DuBoff et al., 2012; Manczak and Reddy, 2012). In ARSACS, mitochondria also accumulate around proximal dendrites (Girard et al., 2012) and exhibit reduced DRP1-dependent fission (Bradshaw et al., 2016). We observed occlusion of mitochondria around vimentin bundles (Figure S2B) with no alterations in the actin cytoskeleton (Duncan et al., 2017) (Figure S2C). To assess how these alterations affect mitochondria in neurons, we performed neuronal differentiation of SH-SY5Y cells (Shipley et al., 2016). While wild-type (WT) and sacsin KO cells expressed indistinguishable levels of neuronal markers, neurites were fewer and shorter in sacsin KO cells (Figures S2D–S2G), contained fewer mitochondria (Figure S2H), and had diminished mitochondrial movement (Figure S2I and Video S3). Our proteomics data also identified several hyperphosphorylated kinesin proteins, which shuttle mitochondria along microtubule tracts (Frederick and Shaw, 2007) (Table S1). In all, these results demonstrate that the loss of sacsin affects microtubule structure, dynamics, and function, in agreement with recent findings demonstrating that sacsin directly interacts with microtubules (Francis et al., 2022). Video S3. Representative live cell spinning-disk confocal microscopy of mitochondria in neuronally differentiated SH-SY5Y cells (WT left, KO right)
To more systematically characterize our proteomic datasets, we performed gene ontology (GO) analysis for the total proteome and phosphoproteome (Figures 3A and 3B; Table S2). The top associated terms in the proteome were related to “focal adhesions,” including “integrin signaling,” “actin filament,” and “regulation of protein localization to plasma membrane.” “Focal adhesion” was also a top term in phosphoproteome, suggesting that FA proteins are affected at both the total protein and post-translational levels. FAs are plasma-membrane-associated macromolecular assemblies that physically link the intracellular cytoskeleton and extracellular matrix (ECM). FAs are composed of integrin receptors bridging the ECM with actin bundles, which interact with microtubules and intermediate filaments to coordinate dynamic regulation of FA structure (Ezratty et al., 2005; Leube et al., 2015; Seetharaman and Etienne-Manneville, 2019). In the brain, FAs are critical for structural remodeling during axon growth, synapse formation, and maintenance (Kilinc, 2018). Immunolabeling for the core FA proteins paxillin and vinculin revealed decreased FA number, area, and aspect ratio in sacsin KO cells (Figures 3C and S3A–S3G) while total levels of these proteins were unaffected (Figure S3H and Table S1). While paxillin is primarily localized at FAs, it also is known to interact with the MTOC (Robertson and Ostergaard, 2011), and we observed perinuclear accumulation of paxillin coinciding with the vimentin bundle (Figure S3A). Microtubules regulate vinculin localization to FAs (Ng et al., 2014), and we found reduced vinculin and vimentin dynamics in sacsin KO cells using fluorescence recovery after photobleaching (FRAP) (Figures 3D, S3I, and S3J). We next removed cell bodies with hypotonic shock, leaving only the structural remnants of cell-ECM interactions, and again found reduced vinculin structures, suggesting that the mislocalization of adhesion proteins also results in decreased cell-ECM interactions (Figures S3K–S3N). These findings were consistent in sacsin KO HEK293 cells, which were generated using an alternative CRISPR-Cas9 genome-editing strategy (Duncan et al., 2017) (Figures S3O–S3S). Our proteomics data also revealed decreased levels of several integrin proteins (Figure S3T). Localization of ITGAV to FAs was diminished in sacsin KO cells (Figure 3E), while ITGA6 was sequestered in the vimentin bundle (Figure S3U). In all, these data suggest that the trafficking, structure, and function of multiple FA proteins is affected in sacsin KO cells. To determine whether levels of adhesion proteins were also affected in neurons, we performed quantitative proteomics of primary cortical neuron cultures derived from embryonic day 15.5 Sacs(−/−) mice (Figure S3V and Table S1). These cultures are composed of ∼75% NeuN+ neurons, but also contain a smattering of other cell types such as astrocytes (Pearson et al., 2016). While no proteins passed statistical cutoffs in both datasets (p < 0.05, log2 fold change ±0.4), comparing levels of proteins which were significantly affected in either dataset revealed a statistically significant relationship, suggesting that a subset of proteins is affected in both cellular contexts (Figure 3F). Notably, these proteins included vimentin and several integrins. The most differentially expressed proteins in cortical cultures included neuron-specific proteins, such as Nrsn1, which binds tubulin and plays a role in vesicular trafficking (Ida et al., 2004; Kiyonaga-Endou et al., 2016), and astrocyte-specific proteins, such as the intermediate filament protein GFAP (Murtinheira et al., 2022). This may suggest that multiple cell types are affected by the loss of sacsin. When we analyzed statistically significant proteins by GO term analysis, we saw similar processes as those identified in SH-SY5Y cells (Figure 3A), including “cell adhesion molecule binding,” “actin filament bundle,” and “focal adhesion” (Figure S3W). As the proteins input into each GO term analysis were completely non-overlapping, this suggests that the loss of sacsin affects cytoskeletal and FA structures independent of cellular context.
Beyond providing structural support for cells, FAs are enriched with many signaling proteins, which transmit signals from the extracellular milieu to effectors in the cytoplasm and nucleus. A master regulator of FA signaling is the FA kinase (FAK/PTK2) (Sulzmaier et al., 2014). FAK is recruited to integrin adhesion complexes through interactions with paxillin (Brown et al., 1996) and is activated via autophosphorylation at Tyr397 following integrin receptor binding to the ECM (Zhao and Guan, 2011). FAK regulates neuronal outgrowth and synapse formation by phosphorylating multiple downstream effectors of FA signaling (Rico et al., 2004) (Figure 4A). Although total levels of FAK were unaltered in sacsin KO cells, pFAK was significantly reduced, as was its localization to FAs (Figures 4B, 4C, S4A, and S4B). JNK and paxillin, downstream targets of activated pFAK (Zhao and Guan, 2011), were also hypophosphorylated, without corresponding changes in protein levels (Figures 4B and S4C–S4G; Table S1). These data suggest that FAK signaling is suppressed in sacsin KO cells, possibly through disengagement with FAs. We next considered the mechanism by which FAK signaling is suppressed in sacsin KO cells. The phosphatase PTEN, which dephosphorylates FAK and negatively regulates FAK activity (Tamura et al., 1999), was elevated in sacsin KO cells (Figures 4B and S4H). To investigate whether increased PTEN is a general consequence of intermediate filament disorganization, we treated WT SH-SY5Y cells with simvastin (Trogden et al., 2018), which induced vimentin bundling and perinuclear accumulation but did not affect PTEN levels (Figures S4I–S4K). Conversely, reducing PTEN by small interfering RNA (siRNA)-mediated knockdown to WT levels in sacsin KO cells (Figures 4D and 4E) increased pFAK and pPAX (Figures 4D, 4F, and 4G), reduced the frequency of perinuclear vimentin accumulation, and increased the number of FAs (Figures 4H–4J). FAs also play an important role in the migratory behaviors of cells (De Pascalis and Etienne-Manneville, 2017), and sacsin KO cells exhibited migration deficits in scratch and transwell migration assays (Figures S4L–S4O), which were rescued by PTEN knockdown (Figures S4P and S4Q). Together these results indicate that increased PTEN activity contributes, at least in part, to the intermediate filament and FA phenotypes in sacsin KO SH-SY5Y cells.
FAs act as signal transduction hubs to integrate information from the outside of the cell to the inside. Some FA proteins, including paxillin and zyxin (Figure 1C), can shuttle to the nucleus and function as transcriptional co-regulators in a phosphorylation-dependent manner (Dong et al., 2009; Suresh Babu et al., 2012). Interestingly, GO term analysis for proteins with altered phosphorylation were highly enriched for terms related to RNA processing, including “RNA binding,” “cytoplasmic stress granules,” “spliceosome,” and “nuclear body” (Figure 3B and Table S2), suggesting that the altered phosphorylation landscape may be affecting the transcriptome. Therefore, we next performed RNA sequencing (RNA-seq) of neuronally differentiated SH-SY5Y cells (Figure S5A and Table S3). We found 876 differentially expressed genes (false discovery rate [FDR] <0.05, log2 fold change ±0.4), suggesting that the loss of sacsin has profound effects on the transcriptome (Figure S5A). Protein interaction mapping revealed altered expression of multiple ECM proteins, integrins, and regulators of integrin activation (Figure S5B). Interestingly, changing the total levels or activity of specific integrins can affect the expression of other integrin subunits, a phenomenon called “integrin crosstalk” (Samarzija et al., 2020). The observation that multiple integrins were affected at both the protein and RNA levels suggests that altered integrin localization may activate regulatory feedback loops which affect the expression of genes that play a role in membrane-based signaling. Indeed, GO term analysis of differentially expressed genes identified terms implicating membrane-related processes, including “postsynaptic membrane,” “axon terminus,” “endomembrane system,” and “cytoplasmic vesicle membrane” (Figure S5C). In all, these data suggest that the altered phosphorylation landscape in sacsin KO cells affects mRNAs encoding for proteins involved in membrane-related processes. Cell surface proteins are frequently under-represented in proteomics experiments owing to low expression and biochemical properties (Bausch-Fluck et al., 2015). Indeed, while 26% of the genes detected by RNA-seq were detected in the proteome, only 11% of differentially expressed genes (which were enriched for membrane proteins) were detected in the proteome (Figure S5D). Therefore, to better characterize membrane and surface proteins, we incubated live cells with biotin, labeling cellular and exosome membrane/surface proteins, followed by neutravidin purification, and analysis by quantitative mass spectrometry (Nunomura et al., 2005) (Figure 5A and Table S1). This approach identified an additional 870 proteins not in our initial proteomic datasets (Figure S5E). Proteins with altered surface expression in sacsin KO cells included several signaling receptors (FGFR1, LRP4) and GTP-binding proteins involved in signal transduction (GNG2, GNG8) (Figure 5B). Two of the most affected membrane proteins were the synaptic adhesion proteins neuronal cell adhesion molecule (NRCAM) and neurofascin (NFASC), which form a molecular complex and have been linked to movement disorders (Kurolap et al., 2022; Kvarnung et al., 2019; Smigiel et al., 2018) (Figures 5B and 5C). We next compared membrane proteins found in both proteomic and surfaceome datasets, reasoning that conflicting levels between cell surface and total protein levels could reflect improper membrane recycling, precocious membrane localization, or deficits in membrane-bound trafficking. Many proteins with altered surface levels showed no or even opposing change in total protein levels (Figure 5D and Table S1). Among the most mislocalized proteins were synaptic adhesion proteins, including multiple integrins (ITGA1, ITGB1, ITGA3), neuronal cell adhesion molecules (NRCAM, CNTN1, LSAMP), the FA regulator RET/GFRA3 heterodimer, the microtubule-binding protein DCX, and AHNAK, a 700 kDa scaffolding protein with diverse yet poorly understood function (Davis et al., 2015) (Figure 5D). GO term analysis of proteins with altered surface levels suggested deficits in processes related to vesicle packaging and transport (Figure 5E). These included eight exosomal Rab proteins, which were increased in the surfaceome and not affected at the total protein level (Figure S5F and Table S1). Rabs are a diverse family of GTPases that coordinate multiple aspects of membrane protein trafficking, including FA turnover, and integrin endo-/exocytosis (Moreno-Layseca et al., 2019). Specific Rabs also regulate trafficking between the Golgi and the endosomal network (RAB8A, RAB10), bidirectional Golgi/endoplasmic reticulum (ER) trafficking (RAB2A, RAB18), and epidermal growth factor receptor (EGFR) internalization (RAB7A) (Bakker et al., 2017; Galea and Simpson, 2015). Kinome profiling also identified multiple regulators of Rab activity and trafficking, including PIK3R4 and PIK3C3, which regulate PTEN activity through localization to vesicles in a microtubule-dependent fashion (Naguib et al., 2015). To assess trafficking and localization deficits in sacsin KO cells we investigated the localization of the ECM protein fibronectin, which is packaged into vesicles in the ER and Golgi (Kii et al., 2016) and trafficked to the cell periphery along microtubules (Noordstra and Akhmanova, 2017). Fibronectin was not affected in any of our proteomics datasets, allowing us to investigate mislocalization independent of changes in protein level or phosphorylation. In WT HEK293 cells, fibronectin puncta were organized in “chains,” which appear collapsed around the vimentin bundle in sacsin KO cells (Figure 5F). Staining for the ER marker KDEL revealed that fibronectin is retained in the ER in HEK293 and SH-SY5Y sacsin KO cells (Figures 5G and S5G), suggesting that membrane-bound trafficking is affected in sacsin KO cells. We next used Ingenuity Pathway Analysis to assess whether the misregulated cell surface proteins are associated with any pathological conditions. Resoundingly, the terms were associated with disease traits reminiscent of ARSACS, including “movement disorders,” “neurodegeneration,” and “progressive neurological disorder” (Figure 5H). Notably, three of the most mislocalized proteins, NFASC, NRCAM, and CNTN1, form molecular complexes that are important for axon guidance (Pollerberg et al., 2013), maintenance of synapses by astrocytes (Takano et al., 2020), and interactions between Purkinje neuron axons and glia (Bhat et al., 2001). KO mice or humans which harbor mutations in each of these genes develop cerebellar ataxias with features that resemble ARSACS (see discussion).
Cerebellar atrophy is an early clinical feature of ARSACS (Martin et al., 2007; Synofzik et al., 2013). In the ARSACS mouse model, the progressive death of Purkinje neurons begins around postnatal day 90 (P90) (Lariviere et al., 2015) and is well under way by P120 (Figure 6A). To determine whether any of the proteins that were mislocalized in our sacsin KO cell model were also affected in the brain, we focused on mice at P60, which is when behavioral deficits first emerge but prior to Purkinje neuron death (Lariviere et al., 2015). ITGA1, which was among the most mislocalized proteins in sacsin KO cells (Figure 5D), is normally localized in nuclear Cajal bodies and Purkinje axons in Sacs(+/−) mice (Figures 6B and 6C). However, in Sacs(−/−) mice, we observed striking accumulation of ITGA1 in the soma and dendritic trunk (Figures 6B–6D). Axonal swelling near the Purkinje neuron soma is a consistent feature in Sacs(−/−) mice (Lariviere et al., 2015), and we also observed ITGA1 accumulation in these structures (Figure S6A). In contrast, we observed a decrease of ITGA1 in Purkinje neuron axon tracts (Figures 6E–6G), likely reflecting reduced ITGA1 trafficking. Purkinje axons synapse onto neurons in the DCN, which in turn project to multiple brain regions. As the primary output hub of the cerebellum (Ito, 2002), alterations in the Purkinje-DCN circuit have substantial effects on both motor and non-motor processes (Baek et al., 2022; Sathyamurthy et al., 2020), and are observed in multiple neurodegenerative ataxias (Barron et al., 2018; Feng et al., 2022; Walter et al., 2006). We observed striking disorganization of Purkinje neuron synapses in the DCN in Sacs(−/−) mice at P60 (Figure 6H) and P120 (Figures S6B and S6C), in agreement with a previous report (Ady et al., 2018). The number of Purkinje synapses on each DCN neuron was reduced in Sacs(−/−) mice (Figure 6I), while the size of Purkinje axon termini apposed to DCN neurons was substantially increased (Figure 6J). We observed accumulation of ITGA1 in large CALB+ structures in Sacs(−/−) mice, suggesting that while ITGA1 trafficking is not altogether abolished in Sacs(−/−) mice, ITGA1 does accumulate in these pathological swellings (Figures S6D and S6E). Interestingly, we also observed increased ITGA1 staining in the cell bodies of DCN neurons (Figures 6K and S6F) and accumulation of ITGA1 in the large-diameter dendrites of DCN neurons (Figures 6K and 6L). This pattern was similar to the dendritic ITGA1 accumulation seen in Purkinje neurons (Figure 6C), suggesting that altered protein localization is not unique to Purkinje neurons. As DCN neurons project throughout the brain, the physical disruption between Purkinje and DCN neurons suggests that cerebellar output to multiple brain regions may be directly affected in ARSACS.
To identify how the loss of sacsin causes abnormal protein trafficking, we performed quantitative label-free mass spectrometry of proteins which co-immunoprecipitate with endogenous sacsin in WT SH-SY5Y cells. KO cells were also used to control for non-specific protein pull-down. Our analysis identified 96 proteins as putative sacsin interactors, including vimentin and vinculin (Table S4). Immunofluorescence revealed sacsin puncta in and around vinculin-positive FAs (Figures S7A and S7B) and in close proximity to vimentin structures, with sacsin often being between them (Figure 7A). Reciprocal co-immunoprecipitation (co-IP) experiments confirmed interactions between sacsin, vimentin, and vinculin, but the interaction between vimentin and vinculin was dramatically reduced in sacsin KO cells (Figure 7B). NFASC has been reported to interact with vimentin (Sistani et al., 2013), leading us to wonder whether NFASC may also interact with FA proteins. Co-IP experiments identified an interaction between NFASC and vinculin, which was dramatically reduced in sacsin KO cells (Figure 7C). These results suggest that sacsin promotes the formation and/or stabilization of adhesion protein interactions. To identify central proteins which may explain the cellular phenotypes in sacsin KO cells, we performed STRING network analysis (Szklarczyk et al., 2019). We considered all proteins altered in any of our datasets and assessed only high-confidence physical or regulatory interactions. k-Means clustering of network interactions identified three clusters, which highlight complementary pathways by which sacsin contributes to cell structure and signaling (Figure 7D). Central to cluster 1 is the interaction between sacsin and intermediate filament proteins, which interact with a variety of cell surface receptors. Combined with our biochemical experiments, this suggests that the loss of sacsin leads to improper localization of adhesion proteins to the plasma membrane, possibly through decreased protein interactions between intermediate filaments, adaptors, and adhesion proteins. The network also highlighted the microtubule-associated kinase MAST1, which stabilizes PTEN (Valiente et al., 2005) and is protected from proteasomal degradation by the sacsin interactor HSP90B1 (Pan et al., 2019). Cluster 2 is composed of the interaction between sacsin, chaperone network proteins, and microtubules, which in concert regulate membrane protein processing, trafficking, and localization (McClellan et al., 2007). Multiple heat-shock protein (HSP) chaperones were part of the sacsin interactome (Figure 7D), including the marker of ER stress HSPA5/BIP and several HSP90 proteins, which can stabilize FAK, modulate cell migration (Xiong et al., 2014), and regulate microtubules (Quinta et al., 2011). Recent evidence suggests HSP90 is essential for microtubule acetylation (Wu et al., 2020), suggesting that the loss of sacsin may alter microtubule stability via HSP proteins (Figures 2B–E). HSPs also regulate Rab proteins (Chen and Balch, 2006) (cluster 3), which have diverse roles in vesicular trafficking, including PTEN and EGFR trafficking (Shinde and Maddika, 2016). Rabs are highly enriched in synapses, play key roles in endo- and exocytosis, and are linked to many neurodegenerative diseases (Kiral et al., 2018). The increased surface abundance of multiple Rab proteins without corresponding changes in total Rab levels is consistent with the precocious microtubule stability and dynamics we observe in sacsin KO cells. GO term analysis revealed that 65% of sacsin interacting proteins are involved in exosome-related processes, with additional interactors being implicated in unfolded protein binding (HSPs) and FAs (Figure S7C). In all, these results suggest that sacsin plays a direct role in bridging protein quality control systems, microtubule-dependent vesicular transport, and membrane localization of adhesion proteins.
This study identifies sacsin as a central regulator of multiple aspects of cellular structure, including intermediate filament architecture, microtubules, protein trafficking, and FAs. The complex and intertwined relationships between these processes complicates our understanding of their precise pathophysiological relevance, but our results raise some intriguing possibilities. Sacsin possesses a functional J domain, which interacts with HSP70 chaperone proteins (Genest et al., 2019; Parfitt et al., 2009) (Figure 7D). HSPs play a role in ubiquitin-dependent turnover of intermediate filaments (Gavriilidis et al., 2018), and neurofilament bundling in ARSACS neurons can be rescued by HSP expression (Gentil et al., 2019). Sacsin also possesses an ATPase domain with homology to HSP90 proteins. The sacsin interactor HSP90B1 stabilizes FAK (Xiong et al., 2014), suggesting that restoring FAK signaling may rescue intermediate filament structure through HSP activity (Figures 4 and 7D). It is also possible that sacsin transiently interacts with HSP90-regulated kinases, such as FAK (Xiong et al., 2014), and has a more direct role at FAs. HSP70/90 complexes bind to microtubules in an acetylation-dependent fashion (Giustiniani et al., 2009) and interact with hyperphosphorylated tau to increase tau’s interaction with microtubules (Lackie et al., 2017). Since HSPs are known to regulate all of the protein clusters with deficits in sacsin KO cells (Figure 7D), we hypothesize that the interaction between HSPs and sacsin may be an especially critical interaction that is lost in ARSACS. Furthermore, as illustrated by sacsin’s mediation of the interaction between intermediate filaments and FAs, changes in additional as yet uncharacterized protein-protein interactions may explain specific ARSACS phenotypes, such as disrupted autophagy, nuclear morphology, and aberrant localization of mitochondria. Proper localization of synaptic adhesion proteins is critical for neuronal health and is disrupted in many neurodegenerative diseases (Kiral et al., 2018). As cell adhesion proteins, integrins play key roles in modulating axon outgrowth, dendritic arborization, and regulating synaptic structure and function (Park and Goda, 2016). More specifically, multiple integrins and pFAK are localized to dendritic spines in cultured Purkinje neurons, where they regulate spine remodeling (Heintz et al., 2016). However, little is known about the role of ITGA1 in the brain (Murase and Hayashi, 1998), and the lack of a mechanistic connection between ITGA1 localization and the changes to synaptic structure in ARSACS mice is a limitation of our findings. As multiple levels of data suggest that integrins as a class are affected in sacsin KO cells (proteomics, transcriptomics, and surfaceomics), exploring the localization of additional integrin subunits may shed light on this question. Furthermore, integrins are in general most highly expressed during brain development (Nieuwenhuis et al., 2018). Thus, defining when changes in integrin mislocalization and synaptic structure first emerge may yield important insight into the pathomechanistic origins of ARSACS. Our data also suggest that restoring FA signaling by reducing PTEN levels may rescue some cellular deficits in sacsin KO SH-SY5Y cells. PTEN is highly enriched in axons, where it regulates neurite outgrowth, organelle trafficking, and synaptic plasticity (Kreis et al., 2014). Reducing PTEN activity with competitive peptides, small molecules, or genetically have shown therapeutic potential in acute models of axonal injury and stroke (Park et al., 2008; Shabanzadeh et al., 2019) as well as a progressive neurodegenerative tauopathy (Benetatos et al., 2020). As PTEN directly regulates multiple pathways, including PI3K/AKT/mTOR, targeting downstream components of PTEN-dependent regulatory cascades may also have therapeutic potential (Jacobi et al., 2022) and bypass concerns over PTEN’s roles in neurodevelopment and tumor suppression (Skelton et al., 2020). However, as we did not detect evidence of increased PTEN levels in our primary cortical culture proteomic data, determining whether PTEN signaling is hyperactive in Purkinje cells in ARSACS requires further investigation. Nevertheless, as the above examples achieve neuroprotective effects by reducing PTEN activity in multiple neurodegenerative contexts, this approach remains an intriguing strategy. Why do mutations in sacsin, which is expressed throughout the brain, present as a cerebellar ataxia? Proteins whose abundance or localization is altered in sacsin KO cells, and which also cause cerebellar ataxia, could suggest a causal molecular deficiency in ARSACS. The interactions between NFASC, NRCAM, and CNTN1 are critical for brain development, and mutation of each causes phenotypes reminiscent of ARSACS. Cntn1 KO mice have deficits in axon guidance and develop cerebellar ataxia (Berglund et al., 1999). Nrcam KO mice have phenotypes only in lobules 4/5 of the cerebellar vermis (Sakurai et al., 2001), which are also specifically affected in ARSACS (Ady et al., 2018; Lariviere et al., 2015, 2019). Lastly, human mutations in NFASC which selectively remove the 155 kDa glial isoform cause congenital hypotonia, demyelinating neuropathy (as in ARSACS), and severe motor coordination deficits (Smigiel et al., 2018), while mutations of the neuron-specific 186 kDa NFASC isoform cause cerebellar ataxia (Kvarnung et al., 2019). These convergent phenotypes lead us to hypothesize that improper localization of synaptic cell adhesion molecules may be a causal molecular deficiency in ARSACS. In development, if an axon fails to make productive synaptic connections and receive neurotrophic input from nearby cells, molecular cascades are activated, which cause localized pruning of non-productive axonal branches (Dekkers et al., 2013). This process, which initiates at the synapse and advances back toward the cell body, is referred to as the dying back model, and can cause neuronal death (Raff et al., 2002). Although this is a normal mechanism to ensure proper wiring of the nervous system in the face of stochastic errors in axon guidance, this process is co-opted in many neurodegenerative disorders, including amyotrophic lateral sclerosis (Dadon-Nachum et al., 2011), Alzheimer’s disease (Salvadores et al., 2017), Huntington’s disease (Han et al., 2010), Parkinson’s disease (Dauer and Przedborski, 2003), and hereditary spastic paraplegias (Fink, 2013). A common molecular thread across these diseases is microtubule-based axonal transport (Morfini et al., 2009), and many of the proteins implicated in the aforementioned diseases were also identified in this study (e.g., tau, tau kinases, Rabs, synaptic adhesion proteins). This leads us to speculate that the loss of sacsin alters microtubule function, resulting in improper trafficking of synaptic adhesion proteins, deficits in synaptic structure, activation of axonal degeneration, and ultimately Purkinje cell death. A mechanistic exploration of this hypothesis will be necessary for the development of rationally designed therapeutic strategies aimed at delaying or preventing ARSACS progression by restoring synaptic structure and function.
The precise molecular function of sacsin remains elusive, in part because of the difficulty of performing biochemical assays with such a large protein. We attempted to shed light on sacsin’s function by identifying interacting proteins, but an important limitation of co-IP experiments is that many of the interactions may be indirect. Determining which proteins interact directly with sacsin may help clarify the mechanism by which sacsin regulates the processes we describe in this study. While we provide multiple lines of evidence that microtubule-dependent trafficking of membrane proteins is affected in sacsin KO SH-SY5Y cells, and some evidence in primary cortical neuron cultures and cerebellar neurons in the brain, a more systematic exploration of this phenotype in Purkinje neurons is warranted. Specifically, analyzing integrin localization at Purkinje synapses, the activity of PTEN/FAK signaling, and the role of vesicular transport and membrane protein turnover will be necessary to assess the physiological relevance of our findings. Furthermore, of the many cellular phenotypes that have been found in ARSACS, teasing apart which are causal and which are merely a part of neurodegenerative processes is necessary to understand the fundamental role of sacsin in the brain. We posit that exploring the neurodevelopmental aspects of this disease, prior to the onset of the neurodegenerative cascade, may help shed light on this question.
Further information and requests for resources and reagents should be directed to the lead contact, Justin Wolter ([email protected]).
Cell lines generated in this study are freely available from the lead authors upon request.
Human female SH-SY5Y neuroblastoma cells were obtained from the American Type Culture Collection and were grown in 1:1 Dulbecco’s Minimum Eagle Medium (DMEM)/Ham’s F12 medium, plus 10% heat-inactivated fetal bovine serum,100 U/mL penicillin and 100 mg/mL streptomycin. Human female HEK293T cell lines were obtained from the American Type Culture Collection and were grown in DMEM, plus 10% heat-inactivated fetal bovine serum,100 U/mL penicillin and 100 mg/mL streptomycin.
All animal procedures used in this study were approved by the Institutional Animal Care and Use Committee at the University of North Carolina at Chapel Hill. Mice were housed in an AAALAC accredited facility in accordance with the Guide for the Care and Use of Laboratory Animal. All animal procedures were approved by the University of North Carolina at Chapel Hill Institutional Animal Care and Use Committee (IACUC). Sacs(−/−) mice were a kind gift from Dr. Stefan Strack. Sacs(+/−) mice were generated by mating Sacs KO mice with wild type C57BL/6J mice (Jackson Laboratories). Primers for genotyping are as follows: WT allele forward primer: 5’ - GCTGTCAGGGGGAAATCTGATAAAG –3', WT allele reverse primer: 5’ - GCAGCACCTTTAGACAAAAGATTGC –3', Sacs KO allele forward primer: 5’ - CAACCTTGGAGAAACTGTGCCTG – 3', Sacs KO allele reverse primer: 5’ - CACCGACGCCAATCACAAACAC –3'.
Cortices were isolated from E15.5 mouse pups (Simon et al., 2019). Cortices from each animal were dissociated in papain (Pierce, 88285) and DNase (Sigma, D4513) for 30 min at 37°C, and filtered through a 70 μm filter. Each animal was genotyped during dissection using the above PCR primers, and cortices from either WT or KO animals were pooled independent of sex. Cells were plated into 3 wells of poly-D-lysine coated 6 well plates at 1 × 106 cells per well in Neurobasal medium (Life Technologies) containing 5% fetal bovine serum (Invitrogen), B27 (17504–044, Invitrogen), Antibiotic-Antimycotic (15240–062, Invitrogen) and GlutaMAX (35050–061, Invitrogen). On day 3, we performed a full media change, replacing with the above media with serum omitted, and supplemented with the antimitotic FDU to inhibit proliferation of non-neuronal cell types. We performed 50% media changes every third day. Pooled cultures from each litter was considered a single replicate.
A detailed list of antibodies and dilutions used in this study is provided in Table S5.
An SH-SY5Y cell line with the sacsin truncation mutation M783∗ was generated using CRISPR/Cas9. We cloned the SACS targeting guide RNA (gRNA) TTTCATGGCTTAAGATGGTTTGG (PAM sequence underlined) into the p1261_GERETY_U6_BasI_gRNA vector for expression of the gRNA under control of the U6 promoter. The gRNA expression vector was co-transfected with a Cas9 expression vector (hCas9, Addgene # 41815) and a targeting vector with homology arms to introduce the M783∗ mutation along with a puromycin selection cassette (pMCS-SACStrunc-PB:PGKpuroDtk) using Lipofectamine 3000. Puromycin-resistant clones were selected and screened by PCR and sequencing.
SH-SY5Y were grown to ∼80% confluency in 13 T75 flasks for harvesting. SH-SY5Y flasks were placed on ice, washed twice with ice cold PBS and harvested by scraping cells in lysis buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 0.5 % Triton X-100, 1mM EDTA, 1mM EGTA, 10mM NaF, 2.5mM Na3VO4, complete protease inhibitor cocktail (Roche), and phosphatase inhibitor cocktail 2 and 3 (Sigma). Lysates were sonicated by pulsing once at 30% for 10 s and then twice at 40% for 10 s with 10 s rest on ice between each pulse (Branson 150 Sonifier). Lysate was transferred to microcentrifuge tube and spun at 14,000 × g for 10 min at 4°C. Lysate was filtered through a 0.2 μm syringe filter and stored at −80°C until all replicates were collected. Protein concentration was quantified using a Bradford assay. Cell lysates (1mg, n = 3) were acetone precipitated overnight and stored at −20°C. Protein pellets were resuspended in 8M urea, then reduced with 5M DTT at 56°C for 30 min and alkylated with 15mM iodoacetamide in the dark at RT for 45 min. Samples were diluted to 1M urea, then digested overnight with trypsin (Promega) at 1:100 trypsin:protein ratio. Samples were acidified then desalted using C18 desalting spin columns (Pierce). A peptide BCA colorimetric assay (Pierce) was performed and 500μg of each sample was individually labeled with TMT6 reagent (Thermo). After labeling efficiency was confirmed, the TMT6 labeled samples were mixed and desalted using C18 desalting spin column (Pierce). A 100μg aliquot was set aside for global proteome analysis, and was fractionated into 4 fractions using a High pH reversed phase fractionation spin column (Pierce). The rest of the sample (∼3mg) was enriched for phosphopeptides using Ti-MAC magnetic beads (ReSyn Biosciences). The Ti-MAC eluate was fractionated into 3 fractions using High pH reversed phase fractionation spin column. Global proteome and phosphoproteome fractions were dried down via vacuum centrifugation and stored at −80°C until LC/MS/MS analysis.
Multiplexed inhibitor bead (MIB) kinase enrichment was performed as previously described with a slightly modified bead composition (Arend et al., 2017). Cells were lysed in MIB-MS buffer (50 mm HEPES, pH 7.5, 150 mm NaCl, 0.5% Triton X-100, 1 mm EDTA, 1 mm EGTA, 10 mm NaF, Phosphatase Inhibitor Mixture 2 (Sigma, P5726) and 3 (Sigma P0044), and 2.5 mm NaV04 plus Protease Inhibitor Mixture (Roche). Samples were sonicated and clarified by centrifugation at 14,000g and filtered through a 0.2 μm filter (Corning, #431219). Protein was quantified by Bradford assay. Specifically, each sample was applied to an individual 350 μL Poly-Prep chromatography column (Bio-Rad) containing the following immobilized kinase inhibitors: CTx-0294885, PP58, Purvalanol B, UNC2147A, VI-16832, UNC8088A. Proteins were eluted from columns by boiling in elution buffer (100 mm Tris-HCl, pH 6.8, 0.5% SDS, 1% β-mercaptoethanol) for 15 min. Samples were incubated at room temperature for 30 min in the dark. dl-DTT was added to bring the final concentration to 10 mm and samples were incubated at room temperature for in the dark for 5 min. Samples were concentrated to a final volume of ∼100 μL in 10K Amicon Ultra centrifugal concentrators and proteins were purified by methanol chloroform extraction. Samples were re-suspended in 50 mm HEPES, pH 8.0, and digested with sequencing grade porcine trypsin (Promega) overnight at 37°C. Samples were extracted with ethyl acetate 3 times to remove residual detergent, desalted using Pierce C-18 spin columns, and submitted to the UNC Michael Hooker Proteomics Core for LC/MS/MS analysis.
On day 10 post plating lysates from primary cortical cultures were acetone precipitated overnight and stored at −20°C. Protein pellets (500 μg, n = 3 per genotype) were resuspended in 8M urea, then reduced with 5M DTT at 56°C for 30 min and alkylated with 15mM iodoacetamide in the dark at RT for 45 min. Samples were diluted to 1M urea, then digested overnight with trypsin (Promega) at 1:100 trypsin:protein ratio. Samples were acidified then desalted using C18 desalting spin columns (Pierce). A peptide BCA colorimetric assay (Pierce) was performed and 300μg of each sample was individually labeled with TMT6 reagent (Thermo). After labeling efficiency was confirmed, the TMT6 labeled samples were mixed and desalted using C18 desalting spin column (Pierce). The TMT sample (3 mg total) was fractionated using high pH reversed phase HPLC (Mertins et al., 2018). Briefly, the peptide samples were separated by offline high pH reverse-phase HPLC (Agilent 126) and fractionated over a 90 min method, into 96 fractions using an Agilent Zorbax 300 Extend-C18 column (3.5-μm, 4.6 × 250 mm). Mobile phase A containing 4.5 mM ammonium formate (pH 10) in 2% (vol/vol) LC-MS grade acetonitrile, and mobile phase B containing 4.5 mM ammonium formate (pH 10) in 90% (vol/vol) LC-MS grade acetonitrile were used for separation. The 96 resulting fractions were then concatenated in a non-continuous manner into 24 fractions, dried down via vacuum centrifugation and stored at −80°C until LC-MS/MS analysis.
2 × 106 SH-SY5Y WT and sacsin KO cells were each plated in nine 10 cm dishes, and cultured until 95% confluent (∼3 days) (n = 3 per cell line). To identify proteins which purify non-specifically, an additional replicate of WT/KO lines were processed as below, but without the addition of Biocytin hydrazide. Cells were lifted using CellStripper Dissociation Reagent (Corning, #25056CI) for 20 min at 37°C and resuspended in 1X PBS (pH 6.5) + 1.6 mM NalO4 and rotated at 4°C for 20 min in the dark. Cells were washed three times then resuspended in 1X PBS (pH 6.5) + 10mM Aniline + 1mM Biocytin hydrazide and incubated at room temperature for 60 min, then at 4°C for 20 min while rotating. After three PBS washes, cell pellets were resuspended in RIPA, rotated at 4°C for 30 min, and sonicated with 1 s pulses at 20% power for 1 min. To enrich for the labeled surface proteins, cells were centrifuged 15,000 rpm for 10 min at 4°C and supernatant was incubated in washed Neutravidin High-Capacity Resin (ThermoFisher #29204) for one hour at 4°C. Resin was added to gravity column and washed with RIPA, 1X PBS (pH 7.4) + 1M NaCl, Ammonium Bicarbonate (ABC) + 2M Urea then resuspended in ABC + 2M Urea +5 mM tris(2-carboxyethyl)phosphine (TCEP) and incubated at room temperature in the dark at 55°C shaking at 300 rpm for 30 min. Iodoacetamide (IAM) was then added to a final concentration of 11 mM and shaken at room temp for 30 min in the dark. Resin was centrifuged at 500g for 5 min and resuspended in 1 mL ABC + 2M urea containing 20 μg trypsin (Fisher #P8101) to fragment peptides at RT overnight. To desalt, samples were acidified to pH < 2 with 10% trifluoroacetic acid (TFA) in C-18 spin column (ThermoFisher #89873), washed, and resuspended in 40% acetonitrile +0.1% formic acid then dried with vacuum centrifugation and stored at −80°C.
Kinome, proteome, and phosphoproteome were analyzed by LC/MS/MS using an Easy nLC 1200 coupled to a QExactive HF mass spectrometer (Thermo Scientific). Samples were injected onto an Easy Spray PepMap C18 column (75 μm id × 25 cm, 2 μm particle size) (Thermo Scientific) using a 120 min method. The gradient for separation consisted of 5–50% mobile phase B at a 250 nL/min flow rate, where mobile phase A was 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in 80% ACN. The QExactive HF was operated in data-dependent mode where the 15 most intense precursors were selected for subsequent HCD fragmentation. For kinome samples, QExactive HF was operated as previously described (Arend et al., 2017). QExactive HF resolution for the precursor scan (m/z 350–1600) was set to 120,000 with a target value of 3 × 106 ions and a maximum injection time of 100 ms. MS/MS scans resolution was set to 60,000 with a target value of 1 × 105 ions and a maximum injection time of 100 ms. The normalized collision energy was set to 27% for HCD with an isolation window of 1.6 m/z. Dynamic exclusion was set to 30 s, peptide match was set to preferred, and precursors with unknown charge or a charge state of 1 and ≥8 were excluded. For TMT proteome and phosphoproteome samples (each biological replicate analyzed in duplicate), QExactive HF resolution for the precursor scan (m/z 350–1600) was set to 60,000 with a target value of 3 × 106 ions and a maximum injection time of 100 ms. MS/MS scans resolution was set to 60,000 with a target value of 1 × 105 ions and a maximum injection time of 100 ms. Fixed first mass was set to 110 m/z and the normalized collision energy was set to 32% for HCD with an isolation window of 1.2 m/z. Dynamic exclusion was set to 30 s, peptide match was set to preferred, and precursors with unknown charge or a charge state of 1 and ≥8 were excluded. Cell surface samples and the primary cortical culture samples were analyzed by LC-MS/MS using a Thermo Easy nLC 1200 coupled to a Thermo Fusion Lumos mass spectrometer. Samples were injected onto a Thermo PepMap C18 trap column, washed, then loaded onto an Easy Spray PepMap C18 analytical column (75 μm id × 25 cm, 2 μm particle size) (ThermoFisher). The samples were separated over a 120 min method, where the gradient for separation consisted of 5–45% mobile phase B at a 250 nL/min flow rate; mobile phase A was 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in 80% acetonitrile. For the cell surface samples, MS1 orbitrap scans were collected at a resolution of 120,000 and 1e6 AGC target. The MS2 scans were acquired in the Orbitrap at 15,000 resolution, with a 1.25e5 AGC, and 50ms maximum injection using HCD fragmentation with a normalized energy of 30%. Dynamic exclusion was set to 30 s and precursors with unknown charge or a charge state of 1 and ≥8 were excluded. For the primary cortical culture TMT proteome fractions (24 total), the Lumos was operated in SPS-MS3 mode (McAlister et al., 2014), with a 3s cycle time. Resolution for the precursor scan (m/z 400–1500) was set to 120,000 with an AGC target set to standard and a maximum injection time of 50 ms. MS2 scans consisted of CID normalized collision energy (NCE) 32; AGC target set to standard; maximum injection time of 50 ms; isolation window of 0.7 Da. Following MS2 acquisition, MS3 spectra were collected in SPS mode (10 scans per outcome); HCD set to 55; resolution set to 50,000; scan range set to 100–500; AGC target set to 200% with a 100 ms maximum inject time.
Kinome, proteome and phosphoproteome raw data files were analyzed with MaxQuant version 1.6.1.0 and searched against the reviewed human database (downloaded Feb 2017, containing 20,162 entries), using Andromeda within MaxQuant. Enzyme specificity was set to trypsin, up to two missed cleavage sites were allowed, carbamidomethylation of C was set as a fixed modification and oxidation of M and acetyl of N-term were set as variable modifications. For phosphoproteome samples, phosphorylation of S,T,Y was set as a variable modification. For phosphoproteome samples, phosphorylation of S,T,Y was set as a variable modification. For proteome and phosphoproteome samples, TMT6plex of peptide N-termini & K was set as a fixed modification and the quantitation type was set to reporter ion MS2. For kinome label-free quantitation, match between runs was enabled. A 1% FDR was used to filter all data. For kinome data, a minimum of two peptides was required for label-free quantitation using the LFQ intensities. Cell surface proteome raw data files were processed using MaxQuant version 1.6.15.0 and searched against the reviewed human database (downloaded Feb 2020, containing 20,350 entries), using Andromeda within MaxQuant. Enzyme specificity was set to trypsin, up to two missed cleavage sites were allowed, carbamidomethylation of C was set as a fixed modification and oxidation of M and acetyl of N-term were set as variable modifications. A 1% FDR was used to filter all data and match between runs was enabled. A minimum of two peptides was required for label-free quantitation using the LFQ intensities. For all proteomic datasets, proteins with a missing value in one replicate were imputed using the KNN imputation method, proteins with two or more missing values were removed from analysis. Linear Models for Microarray Data (LIMMA) was used to calculate log2 fold change of LFQ intensity and perform statistical analysis (Ritchie et al., 2015). FDR was calculated using Benjamini-Hochberg adjusted p values. For proteins identified in the surfaceome, we annotated them as ‘Membrane” or ‘Exosome’ based on DAVID bioinformatics database. Proteins which were identified in unlabeled controls (no biotin) were removed from further analysis. Proteins with p < 0.05 and log2 fold change of KO/WT ±>0.4 were included in downstream analyses. The Kinome tree was generated on CORAL (Metz et al., 2018).
The Homogenous Time Resolved Fluorescence (HTRF) (Degorce et al., 2009) Tau aggregation kit (Cisbio, MA) was used to determine tau aggregate levels in undifferentiated and differentiated SH-SY5Y lysates using Fluorescence Resonance Energy Transfer (FRET). Cells were scraped in ice-cold PBS and pelleted followed by lysis using buffer provided in the kit. Protein lysates were quantified using Bradford assay. Serially diluted cell lysates were tested in duplicate in a 96 well to determinate the optimal concentration of protein lysates. Anti Tau-d2 and anti tau-Tb conjugates were diluted 1:50 from stock to a final volume, which was calculated based on the total number of samples. The two conjugates were diluted separately, mixed at equal ratios and vortexed before 10μL were added to 10μL (10μg total concentration) of the protein lysates in each well. Lysates and conjugates were then incubated for 2h at room temperature and the plate was read on the CLARIOstar plate reader using the HTRF filter cube which allows for sequential detection of Donor and Acceptor fluorescence. Signal was measured as the peak ratio of 655nm (acceptor fluorescence) to 620nm (donor fluorescence). The results of the two emission signals were plotted as HTRF Ratio or DeltaF% values.
For GO term graphs, the list of significant genes for each proteomic experiment were input into geneontology.com with human genome as the background, Fisher’s exact test with FDR correction (Gene Ontology, 2021). Graphs include terms in all categories (biological processes, molecular function, cellular component). Due to the hierarchical nature of GO terms in Panther (i.e. groups of terms have a nested nature to assign relationships between them) we only considered the most proximal term in each hierarchy to ensure terms were specific and directly comparable. These proximal terms are listed as “PARENT” in Table S2. All terms underneath each parent are also listed for reference. Terms were ranked by FDR value and the top ten non-redundant top terms were included in each figure. All terms, including graphs, were generated using R scripts from bio-protocol.org (https://doi.org/10.21769/BioProtoc.3429).
Cells were lysed with RIPA buffer and 20 ug protein was loaded per lane and on a Novex 4–12% (4–20% for BRSK2) Tris-Glycine gel (ThermoFisher). Protein was transferred to PVDF and blocked with 1X Blocker BSA (ThermoFisher). Blots were washed and incubated with primary antibodies followed by HRP secondary antibodies (ThermoFisher). Protein was quantified using ImageJ software. Each lane was normalized to the relative density of GAPDH/ACTB. Cell surface protein isolation (Figure 5A) was validated through western blot by collecting 250 μL resin (bound to cell surface proteins) from the wash column and centrifuging at 2000xg for 5 min to pellet resin and remove residual 50mM ABC + 2M Urea. Resin was then resuspended in 250 μL 2X Laemmli buffer with 5% 2-mercaptoethanol. The control sample contained 10 μL of total cell lysates (collected prior to resin addition) from SH-SY5Y WT and sacsin KO cells. The samples were run on a 4–15% TGX Bio-Rad pre-stained gel (Bio-Rad #4568094) and transferred to a PVDF membrane, which was blocked in 5% milk in 1X TBST. Primary antibodies were diluted 1:1000 in 5% BSA in 1X TBST. Goat anti-rabbit HRP secondary antibody was diluted 1:5000 in 5% milk. Remaining immunoblotting was performed as described previously (Duncan et al., 2017).
SH-SY5Y WT and sacsin KO cells were fractionated with the Minute Plasma Membrane Protein Isolation and Cell Fractionation Kit (Invent). The cytoplasmic fraction was lysed in Buffer A and the plasma membrane fraction was lysed in RIPA buffer, both fractions contained protease inhibitor (Pierce). Proteins were diluted 1:1 with 2× laemmli buffer with betamercaptoethanol and run on TGX prestained gels (Biorad). Total protein images were obtained before transferring onto a PVDF membrane. Individual protein levels were normalized to total protein image using Image J software for quantification.
The cleavable, homo-bifunctional cross-linker dithiobis[succinimidylpropionate] (DSP; Pierce, Rockford, IL, USA) was diluted to a final concentration of 1mM in PBS and added to the cultured cells. After incubation for 1h at room temperature, cross-linking was stopped by addition of Tris (pH 7.5) to a final concentration of 20mM. Cells were then washed twice in ice-cold PBS, before the cells were harvested in RIPA buffer (NFASC IP) or 50mM Tris-HCl pH7.4 + 150mM NaCl + 1mM EDTA +0.5% Triton X-100, supplemented with protease inhibitors (Pierce/Roche) and incubated on ice for 5 min (sacsin/VIM IP).
A small aliquot of supernatant was removed for analysis by immunoblotting (input fraction) the remaining supernatant was incubated with rabbit monoclonal anti-vimentin antibody or rabbit monoclonal anti-sacsin overnight at 4°C on a rotor. After 16 h 50μL of magnetic beads (Sigma, Poole, UK) were washed twice in PBS-Tween 0.1% buffer, before being recovered in a magnetic separator. The beads were then resuspended within the cell lysate already incubated with the antibody for 2h at room temperature and washed.
Protein concentration was assessed with a Bradford assay and a small aliquot of supernatant was removed for analysis by immunoblotting (input fraction). 300 μg of the sample were used for the immunoprecipitation. First, 20 μL of protein A magnetic beads (73778, CST) washed with 1× Cell Lysis buffer (9803, CST) were added to the samples for 20 min for preclearing. Next, beads were removed and 5 μL of the precipitation antibody, anti-neurofascin (PA5-78668, Invitrogen) or normal rabbit IgG isotype control (2729, CST), were added to the samples and incubated overnight at 4°C on a rotor. After 16 h, fresh prewashed magnetic beads were added to the samples and rotated at room temperature for 20 min before being recovered in a magnetic separator and washed five times with 1× Cell Lysis Buffer.
All cell culture samples, with the exception of samples prepared for microtubule staining, were fixed with 4% PFA and permeabilized with 0.1 to 0.3% Triton X-100. For microtubule staining, samples were fixed with 100% methanol for 20 min at −20°C. Non-specific binding was reduced by blocking in 5% Normal donkey serum or 1% bovine serum albumin (BSA)+10% normal goat serum. Cells were incubated with primary antibodies for a minimum of 1 h at RT, followed by incubation with labelled secondary antibodies, Phalloidin, and Hoechst for 1 h at RT. Samples were mounted either in Fluoro-Gel mounting medium (Electron Microscopy Sciences, Cat. # 17985–30) or ProLong Diamond Antifade Mountant (Invitrogen, Cat. #P36961). Widefield images were captured with GE IN Cell 2200 high content imaging system equipped with a Plan Fluor 20×/0.75 NA air objective. Confocal images were acquired using an inverted Olympus FV3000RS. Plan Apo 60×/1.4 NA oil (Olympus) or Plan Apo 30×/1.05 NA silicon oil (Olympus) objectives were used. SIM images were acquired and reconstructed in 3D-SIM mode using a Nikon N-SIM system equipped with a Plan Apo TIRF 100×/1.49 NA oil objective. For FA immunolabelling, coverslips were coated with 10μg/mL fibronectin solutions overnight. Confocal microscopy was performed using a LSM880 (Zeiss) with a 63× objective and an AiryScan module. Quantification of incidence of cells with perinuclear vimentin accumulation and incidence of FA was performed blind to experimental status. Imaging processing was carried out with Zen Blue software (Zeiss). FA isolation from cells was performed by hypotonic shock to remove cells while leaving FAs intact, as described previously (Kuo et al., 2012). Isolated FAs were then immunolabelled to detect vinculin and analysed using confocal microscope with quantification as described previously. All image processing steps were carried out using ImageJ software.
For Total Internal Reflection Fluorescence (TIRF) imaging, SH-SY5Y cells were cultured on 35 mm glass bottom dishes (Mattek) and transfected with plasmid expressing EB1-GFP using FuGene transfection reagent (Promega). pGFP-EB1 was a gift from Lynne Cassimeris (Addgene plasmid #17234). TIRF live cell imaging was carried out 24 h after transfection on an inverted Nikon Eclipse Ti2 equipped with a Plan Apo TIRF 100×/1.49 NA oil (Nikon) objective. TIRF images were captured at single z-plane, every second for a period of 1 min. Tracking of EB1-GFP comet was performed using the FIJI plugin TrackMate with the following analysis settings: Laplacian of Gaussian detector with an estimated spot diameter of 0.16 μm, subpixel localization enabled, simple LAP tracker, minimum number of spots on track >9, and maximum number of spots on track <35. Each trajectory was visually inspected to confirm tracking accuracy. Mean track velocities were plotted and statistical analyses were performed using GraphPad Prism 9.0.
SH-SY5Y cells were treated with 10 μM nocodazole (Sigma, Cat. #SML1665) or DMSO control for 3 h at 37°C. Washout of nocodazole was performed by one wash with ice old PBS and one wash with pre-warmed media. Cells were fixed with 100% methanol at indicated timepoints. Measurement of microtubule staining in untreated cells and repolymerization after nocodazole washout were performed on images constructed with maximum intensity projections. Areas occupied by microtubules and cells were thresholded and quantified using a custom CellProfiler pipeline. Microtubule density was quantified by thresholding and normalizing area occupied by microtubules to cell area.
SH-SY5Y cells were differentiated using a modified version of a previously described method (Shipley et al., 2016). Cells were grown to near confluence under normal maintenance conditions prior to the start of differentiation. 1 × 105 cells/well were seeded in a 6 well plate. After 24 h media was changed to DMEM containing 2% Fetal Bovine Serum (FBS: Gibco, #A3840001) and10 μM of all-trans retinoic acid (ATRA: Sigma, #R2625). At day 4 cells were passaged and resuspended in neurobasal media (Gibco, #21103049) supplemented with B27 (Gibco, #17504044) 50 ng/mL Brain-Derived Neurotrophic Factor (BDNF: Sigma, #B3795), and 10 μM RA. At day 12 media was supplemented with 2 mM dibutyryl cyclic AMP (dc-AMP: Santa Cruz, sc-201567A). Media was changed every other day with fresh aliquots of RA, BDNF and dcAMP. Cells were harvested at day 15 for downstream analysis.
Light sheet fluorescence microscopy was used to image mitochondrial movement. Cells were incubated with 100 nM of MitoTracker Green FM (Invitrogen, Cat. #M7514) for 20 min at 37°C and then washed with prewarmed media prior to imaging. Mitochondrial movements along the neurites were captured every at 0.5 μm z-steps, 10 s intervals, for 3 min on a ASI’s RAMM frame (ASI), equipped with Mizar TILT light sheet illumination module (Mizar Imaging LLC), Prime 95B sCMOS (Photometrics) camera, and a Plan Apo TIRF 100×/1.45 NA oil (Nikon) objective. Prior to fluorescence imaging, a brightfield image was captured to confirm presence of neurites. Kymographs of mitochondrial movement along neurites were generated using the line tools and resliced command in FIJI.
Plasmids encoding tdTomato:Vinculin (Addgene #58146) or vimentin-EGFP (Addgene #56439) were transfected with Lipofectamine 3000 in Opti-Mem (ThermoFisher) according to the manufacturer’s instructions. To induce vimentin bundle cells were treated for 4 or 24 h with 10 μm simvastatin, in their standard culture medium. For PTEN knockdown a combination of four siRNAs targeting PTEN were used (ON-TARGETplus Human PTEN [5728] siRNA - SMARTpool: Target sequence1#: Sense GAUCAGCAUACACAAAUUA, Antisense: UAAUUUGUGUAUGCUGAUC; Target sequence2#: Sense GACUUAGACUUGACCUAUA, Antisense: UAUAGGUCAAGUCUAAGUC; Target sequence3#: Sense GAUCUUGACCAAUGGCUAA, Antisense: UUAGCCAUUGGUCAAGAUC; Target sequence4#: Sense CGAUAGCAUUUGCAGUAUA, Antisense: UAUACUGCAAAUGCUAUCG). These siRNAs were at a concentration of 10 nm each and were transfected in combination using Lipofectamine 3000 (ThermoFisher), according to the manufacturer’s instructions. A negative control siRNA with no significant sequence similarity to human gene sequences was used as a control.
Cells were transfected with plasmid encoding vimentin-EGFP or tdTomato:Vinculin. 48 h post transfection FRAP experiments were conducted as described previously using an LSM880 microscope (Zeiss) (Girard et al., 2012). 2μm x 2μm regions of interest in cells expressing vimentin-EGFP or tdTomato:Vinculin were excited with the 488 nm or 568nm laser lines respectively. Following photobleaching regions of interest were imaged a minimum of 50 cycles with a 1 s interval. Image and data acquisition were performed using Zen Black and Zen Blue software (Zeiss). Fluorescence intensity in the photo-bleached regions at each time point were quantified as a percentage of fluorescence intensity before photobleaching.
Cell migration and invasion abilities were assessed using Transwell cell culture inserts (BD Biosciences). For the cell migration assay, 2.5 ×104 cells in 500 μL in serum-free medium were seeded directly into the wells of Transwell chambers with 8 μm-pore membranes. Medium containing 10% FBS, was added into the lower chamber. After 24h, cells were fixed and stained with 2% Giemsa blue stain (Sigma). Cells adhering to the upper surface of the membrane were removed using a cotton applicator. Cells on the lower side of the membrane were counted. Five fields were randomly selected per cell line and the mean number of cells quantified.
Cells were detached from the tissue culture plate using 0.25% Trypsin-EDTA solution and plated at the appropriate number of cells in a 6-well plate for 100% confluence in 24 h. In a sterile environment the monolayer was scratch with a pipette tip forming a 1-2mm scratch from one edge of the well to the other. The media was removed and replaced with 2mL of fresh media. Following the generation and inspection of the wound the plate was placed in an incubator set at 37°C and 5% CO2. Pictures were taken at different time points. A time-lapse microscope with a controlled temperature at 37°C and 5% CO2, was also used in parallel experiments.
Mice were anesthetized using pentobarbital and perfused with 4% PFA in PBS via transcardiac perfusion. Tissue was dissected and drop-fixed in 4% PFA for 24 h at 4°C, followed by a 48-h incubation in 30% sucrose at 4°C. Tissue was mounted in M1 Embedding matrix (ThermoFisher) and stored at −80°C. Tissue was sectioned on a CryoStar NX50 (ThermoFisher) in 40μm sections. Slices were transferred into 200μL of permeabilization buffer (5% NDS, 0.3% Triton X-100, 2% DMSO, 0.01% Sodium Azide, 1X PBS) for 60 min on a shaker at room temperature. Following the incubation, buffer was removed and 200μL of primary antibodies diluted in staining buffer were added for 24–48 h with gentle rocking at room temperature. Sections were washed with PBST (0.3% Triton X-100/1X PBS) three times for 10 min, followed by 2 h room temperature incubation with gentle rocking in the dark in 100 μL staining buffer with secondary antibodies. The secondary antibody was removed and DAPI diluted to 1:1000 was added for 5 min before 3 washes in PBST (0.1% Triton X-100/1X PBS). Slices were mounted on Superfrost Plus slides (ThermoFisher), dried, and ∼200μL mounting medium (Sigma, Polyvinyl alcohol mounting medium with DABCO®, antifading) was added prior to placing a cover slip. Slides were dried overnight prior to imaging. Images were acquired on the Nikon 710 confocal microscope.
Raw image Z-stacks were visualized in FIJI. To quantify Calb+ synapses onto DCN Neun+ neurons (Figures 6I and 6J), we selected individual DCN neurons that were ∼20–25 um in diameter using only the NEUN channel (being blind to CALB1 staining). For each DCN neuron the single Z-stack at the widest diameter was manually isolated. Each DCN neuron was then analyzed using a custom CellProfiler pipeline. In brief, we measured mean intensity and size of Calb+ objects within 1 μm of each DCN neuron. n = 4 animals for each genotype, two sections per animal, for a total of 137 DCN neurons. Images in Figures 6L and S6F were quantified using a custom CellProfiler pipeline. In brief, soma and axons were isolated based on their sizes, and mean intensity was calculated for each object.
Proteins were extracted from 8.5 × 106 cells of KO and WT SH-SY5Y cell lines using 500 μL of the lysis buffer (RIPA with phosphatase and protease inhibitors (EDTA free, Merk). The lysates were cleared by centrifugation (13000g) for 15 min at 4°C and then incubated with 1/50 dilution of anti-sacsin antibody (Abcam, ab181190) overnight at 4°C with slow rotation. Protein A Dynabeads (Merk) were equilibrated with the ice-cold lysis buffer and incubated with each cell lysate/antibody mix for 2 h at 4°C with slow rotation. The beads were then washed with the ice-cold lysis buffer and the bound proteins were eluted with 40μL of 2xLaemmli buffer. The samples were then heated at 95°C for 5 min, centrifuged and loaded onto the NuPAGE™ 4 to 12%, Bis-Tris, 1.5 mm, Mini Protein Gel (ThermoFisher). The gels were resolved in 1× MOPC for 2 h at 4°C and then stained with SimplyBlue SafeStain (ThermoFisher) according to the manufacturer’s protocol. Each gel lane was then cut into 10 fragments and the proteins were extracted from the gel using trypsin digest protocol as previously described (Patel et al., 2009). Digests were analyzed using a Waters NanoAcquity Ultra-Performance Liquid Chromatography system and data processed using PLGS v3.0.2 (Waters, UK). For protein interactome analysis we only focused on proteins which were completely absent from all KO lysates, as these are the most stringent, high-confidence interactors. However, several proteins were identified in both WT/KO lysates, but had substantially reduced intensity values in KO cells. These proteins are also included in Table S4.
For each experiment, the statistical test used, sample size, definition of a replicate, and precision measures are defined in the corresponding figure legend. Center points in all figures are mean of all replicates. Statistical tests were performed using either R or Excel. |
|
PMC9647045 | Harry Alexopoulos,Ioannis P Trougakos,Meletios-Athanasios Dimopoulos,Evangelos Terpos | Clinical usefulness of testing for severe acute respiratory syndrome coronavirus 2 antibodies | 10-11-2022 | In the COVID-19 pandemic era, antibody testing against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has proven an invaluable tool and herein we highlight some of the most useful clinical and/or epidemiological applications of humoral immune responses recording. Anti-spike circulating IgGs and SARS-CoV-2 neutralizing antibodies can serve as predictors of disease progression or disease prevention, whereas anti-nucleocapsid antibodies can help distinguishing infection from vaccination. Also, in the era of immunotherapies we address the validity of anti-SARS-CoV-2 antibody monitoring post-infection and/or vaccination following therapies with the popular anti-CD20 monoclonals, as well as in the context of various cancers or autoimmune conditions such as rheumatoid arthritis and multiple sclerosis. Additional crucial applications include population immunosurveillance, either at the general population or at specific communities such as health workers. Finally, we discuss how testing of antibodies in cerebrospinal fluid can inform us on the neurological complications that often accompany COVID-19. | Clinical usefulness of testing for severe acute respiratory syndrome coronavirus 2 antibodies
In the COVID-19 pandemic era, antibody testing against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has proven an invaluable tool and herein we highlight some of the most useful clinical and/or epidemiological applications of humoral immune responses recording. Anti-spike circulating IgGs and SARS-CoV-2 neutralizing antibodies can serve as predictors of disease progression or disease prevention, whereas anti-nucleocapsid antibodies can help distinguishing infection from vaccination. Also, in the era of immunotherapies we address the validity of anti-SARS-CoV-2 antibody monitoring post-infection and/or vaccination following therapies with the popular anti-CD20 monoclonals, as well as in the context of various cancers or autoimmune conditions such as rheumatoid arthritis and multiple sclerosis. Additional crucial applications include population immunosurveillance, either at the general population or at specific communities such as health workers. Finally, we discuss how testing of antibodies in cerebrospinal fluid can inform us on the neurological complications that often accompany COVID-19.
The infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) triggers both B-cell and T-cell responses directed against all viral antigens including the Nucleocapsid (N) and Spike (S) proteins. S protein is essential for viral entry into host cells and N protein is the most abundantly expressed immunodominant protein. Following the initial contact with the virus and fueled by pro-inflammatory cytokines an antibody response is mounted. Specific antibody tests can reliably detect the presence of these antibodies in biological fluids including serum, plasma, saliva [1], urine, human milk [2] and cerebrospinal fluid (CSF) [3]. Depending on the type and timing of test as well as on the kind of fluid tested different clinical information can be procured. In this review article, we will succinctly describe some of the clinical applications of anti-SARS-CoV-2 antibody testing with a view to the future of the still evolving COVID-19 pandemic. Several studies indicate that most immunocompetent persons develop an adaptive immune response following contact with the virus, irrespectively of disease severity. Antibodies including those of the IgM, IgA and IgG classes against N and S proteins can be detected in the serum as early as 1–3 weeks post-infection, whereas IgM decay rapidly, IgG and IgA can persist for several months (Fig. 1 ). Nevertheless, the titer and exact duration of the anti-SARS-CoV-2 antibodies persistence in the circulation after the clearance of the infection varies and is likely donor-specific while it also depends on disease severity [4,5]. Specifically, antibody titers in most cases correlate with disease severity as subjects with more severe COVID-19 raise higher titers (see below) and exhibit longer persistence [6]. Different assays can be used to a. measure the titer of antibodies and their binding to specific SARS-CoV-2 antigens (e.g., N or S proteins) and/or b. determine their specific neutralizing activity. Binding tests fall into two broad categories. Point-of-care tests are performed in any setting e.g., hospital ward, nursing home or workplace and they usually are lateral flow devices that detect antibodies in a blood drop. Laboratory tests require specialized personnel and include methods such as ELISA (Fig. 2 ) and chemiluminescence assays (CIA/CLIA) that detect the antibodies from serum, plasma, dried blood spots or CSF (Fig. 2). In total 85 tests have received EUA for serology from the FDA (https://www.fda.gov/medical-devices/coronavirus-disease-2019-covid-19-emergency-use-authorizations-medical-devices/in-vitro-diagnostics-euas-serology-and-other-adaptive-immune-response-tests-sars-cov-2). This approval categorizes laboratories in three categories, namely H for meeting requirements to perform high complexity tests, M for meeting requirements to perform moderate complexity tests and W which are patient care settings. Neutralizing assays are a proxy of the capacity of antibodies against S to block viral binding to its cognate receptor i.e., the angiotensin converting enzyme 2 (ACE2) and thus entry in human cells (Fig. 2). Types of tests in this category include virus neutralization, pseudo-virus neutralization and competitive neutralization [7]. The first two types require more time, specialized personnel and facilities for handling pathogens, while the latter (in a plate format) is commercially available and easy to set-up and perform in a standard wet lab. Currently, only two tests have received EAU from the FDA (https://www.fda.gov/medical-devices/coronavirus-disease-2019-covid-19-emergency-use-authorizations-medical-devices/in-vitro-diagnostics-euas-serology-and-other-adaptive-immune-response-tests-sars-cov-2). Serological assays that can reveal humoral immune responses against SARS-CoV-2 play a critical role in informing public health providers as well as in directing health care decisions and policies. Currently, antibody tests are mostly performed in central clinical laboratories with a limiting broad access to diverse populations. Moreover, it is important to provide highly sensitive assays that can discriminate between SARS-CoV-2 infection and vaccination. To this end various novel methods are still under development including a multiplexed nano plasmonic biosensor [8] or a microfluidic cartridge based devise [9]; these can be deployed as point-of-care (PoC) antibody profiling methods even in non-specialized places such as the workplace or a nursing home.
Several clinical applications can be informed by antibody testing after careful consideration of the analyte measured (type of Ig and/or kind of antigen, see Table 1 ) and the biological fluid used.
Antibody testing against any viral antigen is not as sensitive and (in some cases) specific as molecular (i.e., PCR detection of SARS-CoV-2 genes) testing. Nevertheless, antibody testing can be for instance useful to identify infected subjects in vaccinated populations. To this end, specialized testing is being developed to simultaneously detect antibodies against S, RBD and N proteins and even neutralizing capacity of antibodies [10] in the blood or in saliva. SARS-CoV-2 antibodies in saliva serve as first line of defense against the virus. They are present in the mucosa, more precisely in saliva, after a recovered infection. Reportedly, antibody persistence in plasma and in saliva was shown in up to 15 months after mild COVID-19 [11]. Notably, salivary IgA and IgG antibodies could be detected earlier in patients with mild COVID-19 symptoms as compared to severe cases [12]. However, severe COVID-19 triggered higher salivary antibody and blood antibody titers than asymptomatic or mild infections [13]. Salivary IgA titers quickly decreased after 6 weeks in mild cases but remained detectable until at least week 10 after severe COVID-19. In conclusion, assays for both IgA and IgG have high specificity and sensitivity for the confirmation of current or recent SARS-CoV-2 infections and evaluation of the IgA and IgG immune responses [12]. Another novel method for seroprevalence studies employs SARS-CoV-2 IgG FcγR ELISAs, methodically combining antigen-antibody binding in solution and isotype-specific detection of immune complexes, allowing for the long-term detection of anti-SARS-CoV-2 IgG antibodies in populations with a challenging immunological background and/or in populations which S-protein-based vaccine programs have been rolled out [14]. Antibody testing can also be potentially useful for long-COVID detection, a yet poorly defined clinical entity. Specifically, it has been shown that 42–53% of subjects with long COVID, but without detectable SARS-CoV-2 antibodies, nonetheless had detectable SARS-CoV-2 specific T cell responses; these findings demonstrate the diagnostic complexity of long COVID and how is compounded in many patients who were or might have been infected with SARS-CoV-2 but not tested during the acute illness and/or are SARS-CoV-2 antibody negative [15]. Additionally, machine learning approaches that input parameters such as serum pro-inflammatory, anti-inflammatory and anti-viral cytokine and anti-SARS-CoV-2 antibody measurements can also help stratify patients at hospital admittance into high and low risk clinical categories with distinct cytokine and antibody profiles that may guide personalized therapy [16]. Given the ongoing anti-COVID-19 vaccination campaign in many countries globally, monitoring of breakthrough infections is also a matter of great importance. Since COVID-19 vaccines induce immune responses against the S protein, which is the main sero-surveillance target to date, alternative targets are being explored to distinguish infection from vaccination. The sensitivity of N seropositivity was 85% for mild COVID-19 in the first two months following symptoms onset but sensitivity was lower in asymptomatic individuals (67%). N-specific IgG concentrations were not affected by vaccination in infection-naïve participants therefore serological responses to N may prove helpful in identifying SARS-CoV-2 infections post-vaccination [17]. Similarly, in a cohort of adult paramedics in Canada, the performance of N antibodies detection was investigated to identify previous COVID-19 infections and compare differences among vaccinated and unvaccinated donors. It was found that vaccinated and unvaccinated groups require different thresholds to achieve optimal test performance, especially for detecting SARS-CoV-2 infection within the preceding 9 months [18].
Several studies aim to predict COVID-19 severity in unvaccinated individuals by using anti-SARS-CoV-2 serological responses. To this end the kinetics of the serological responses along with the correlation between the antibody titers and disease outcome have been assessed. It has been found that antibody titers gradually increased for up to 3 weeks since the onset of symptoms for patients requiring oxygen supplementation with significantly higher antibody titers for patients requiring invasive ventilation [13]. Antibody titers on admission were also significantly higher in severely ill patients and serology performed well in predicting the necessity of invasive ventilation [19]. Similar results were obtained from another study that showed that high IgG levels against S positively correlated with biomarkers of immune activation and inflammation, while they were negatively correlated with pulmonary function and the extent of pulmonary CT abnormalities. It was thus proposed that S-specific IgG levels serve as a useful immunological surrogate marker for identifying at-risk individuals with persistent pulmonary injury who may require intensive follow-up care after COVID-19 [20]. Regarding protection, in the coronavirus efficacy (COVE) phase 3 clinical trial, vaccine recipients were assessed for neutralizing and binding circulating antibodies as correlates of risk for COVID-19 disease and as correlates of protection. Antibodies were measured at the time of second vaccination and 4 weeks later. It was found that vaccine recipients with post-vaccination 50% neutralization titers 10, 100, and 1000 had estimated vaccine efficacies of 78%, 91% and 96%, respectively, suggesting that measuring neutralizing activity is a strong predictor of vaccine efficacy and can be used to inform vaccination strategies [21]. To apply this concept in the community, i.e., that neutralization is a proxy for actual protection, several efforts are under way to provide lateral flow Point of Care (PoC) tests that can measure levels of RBD-ACE2 neutralizing antibody (NAb) from whole blood, with a result that can be determined by eye or quantitatively on a small instrument [22]. Also significant is the demonstration that such tests can show high correlation with conventional neutralization tests [23]. Convalescent individuals who previously recovered from COVID-19 have enhanced immune responses after vaccination (hybrid immunity) compared with their naïve-vaccinated peers; however, the effects of post-vaccination breakthrough infections on humoral immune response and predicted levels of protection remain to be determined. This was addressed in a study where neutralizing antibody responses were measured in 104 vaccinated individuals, including those with breakthrough infections, hybrid immunity, and no infection history. It was shown that immune sera after breakthrough infection and vaccination after natural infection broadly neutralize SARS-CoV-2 variants to a similar degree. These data suggest that the additional exposure to antigens derived from natural infection substantially boosts the quantity, quality, and breadth of humoral immune response regardless of whether it occurs before or after vaccination [24]. Nevertheless, the molecular details of hybrid immunity warrant additional future studies. Omicron variants share some distinct characteristics that are different from initial SARS-CoV-2 mutants and thus protection of prior infection against reinfection with omicron ranged from 18.0% for patients infected in the first wave of COVID-19 to 69.2% for those infected in the Delta wave [25]. Across the same lines of research, the extent to which Omicron infection, with or without previous vaccination, elicits protection against the previously dominant Delta variant was investigated and it was shown that vaccination combined with Omicron/BA.1 infection hybrid immunity is likely protective against Delta and other variants. In contrast, infection with Omicron/BA.1 alone offered limited cross-VoCs protection despite moderate enhancement [26]. Interestingly, long-term studies (up to 18 months) have showed persistent circulating antibodies even after mild infection, indicating that further work into the detailed immunological mechanisms that govern persistence is required [27]. Also, as mentioned above, higher titers of antibodies have been reported in patients suffering from more severe forms of the disease. These high titers were also detected in deceased patients, compared to asymptomatic patients indicating that fatal infection is not associated with defective humoral response [28].
Healthy individuals respond adequately to full vaccination, although this response is often gender- and age-dependent and is reduced with time [29], [30], [31], [32]. However, either binding or neutralizing antibody responses after vaccination in healthy people is reduced against the different mutants, especially against omicron; therefore the administration of a booster dose is of value [33], [34], [35], [36]. To address the issue of vaccine response in populations receiving immunomodulatory drugs several studies have been conducted to date. The studies discussed below use binding titers against S protein as the key measure, except from a few were we specifically state that a neutralizing assay was used. Methodologically, the use of either ELISA or electrochemiluminescence is of equal value, as both type of methods are FDA-approved. Studies following immunosuppression, include patients with rheumatic diseases, multiple sclerosis and cancer. Reportedly, COVID-19 vaccine-induced antibody responses were altered in patients with inflammatory bowel disease on commonly used immunosuppressive drugs. More specifically, patients on six different immunosuppressive treatment regimens (thiopurines, infliximab, a thiopurine plus infliximab, ustekinumab, vedolizumab, or tofacitinib) were recruited along with healthy control participants from nine centers in the UK. Eligible participants had received two doses of COVID-19 vaccines and antibodies were measured at 53–92 days post-second vaccine dose. The immunogenicity of COVID-19 vaccines varied according to drug, and was attenuated in recipients of infliximab, infliximab plus thiopurines, and tofacitinib [37]. In another prospective observational multicenter study, that included 478 patients with rheumatoid arthritis (RA), systemic lupus erythematosus (systemic sclerosis (SSc), cryoglobulinemic vasculitis and a miscellanea of 13 systemic vasculitis a significantly lower neutralizing antibody response was shown in patients versus controls. Increased prevalence of non-response to vaccine was attributed in those treated with glucocorticoids, mycophenolate-mofetil or rituximab [38]. Also, in patients with autoimmune rheumatic diseases under therapy with methotrexate (MTX) it was shown that MTX reduces the immunogenicity of SARS-CoV-2 vaccination in an age-dependent manner. It was suggested that holding MTX for at least 10 days after vaccination significantly improves the antibody response in patients over 60 years of age [39]. The fact that rituximab (an anti-CD 20 monoclonal antibody) interferes with vaccine efficacy is supported by another study of chronic rituximab treated patients which showed only 36% seroconversion after vaccination [40]. Notably, in ANCA-associated vasculitis patients, and despite the lack of a measurable humoral immune response, B-cell depleted patients mounted a similar vaccine induced antigen-specific T-cell response compared to B-cell recovered patients and normal controls [41]. Whether more vaccine doses can change this outcome was tested in RA patients on rituximab in a prospective, cohort study, where patients with insufficient serological responses to two doses were allotted a third vaccine dose. This third dose given 6–9 months after a rituximab infusion still did not induce a robust serological response, but was considered to boost the cellular T-cells immune response [42]. These studies in patients treated with rituximab clearly call for a re-evaluation of the 6-month interval between treatment and vaccination [43]. Similarly, the antibody levels in multiple sclerosis (MS) patients on anti-CD20 therapy (either rituximab or ocrelizumab), were assessed. It was found that 14.0%, 37.7%, and 33.3% were seropositive after the first, second and third vaccination, while no difference was found in antibody levels after the second and third dose. These findings suggest the need for clinical strategies to allow B cell reconstitution before boosting vaccination [44]. In another multicenter prospective study in MS patients, seroconversion was lowest in patients on anti-CD20 monoclonal antibodies followed by patients on sphingosine-1-phosphate-receptor-modulators [45]. This finding was independently confirmed in two other studies [46], including a large study from Israel where apart from showing that fingolimod- or ocrelizumab-treated patients had diminished humoral responses it was also found that fingolimod compromised cellular immune responses, with no improvement after the third boosting dose. Nevertheless, vaccination following >5 months since ocrelizumab infusion was associated with better seropositivity [47] while CD4+ and CD8+ T-cell responses were preserved [[48], [49]]. In dialysis patients, plasma samples were analyzed from 130 hemodialysis and 13 peritoneal dialysis patients after two doses of BNT162b2 or mRNA-1273 vaccines. It was found that 35% of the patients had low-level or undetectable IgG antibodies to S and that neutralizing antibodies against the vaccine-matched SARS-CoV-2 and Delta were low or undetectable in 49% and 77% of patients, respectively. In these cases, antibody responses must be continuously monitored to adopt the best prophylactic and/or therapeutic strategy [50]. In patients after allogeneic stem cell transplantation vaccination efficacy might be impaired depending on the immune reconstitution. It has been shown that most patients did develop a high antibody titer (138 out 182 patients, 75.8%); while patients undergoing allogeneic stem cell transplantation have been excluded from the initial registration trials, this real-world study showed that most patients do have an adequate response to mRNA vaccines [51]. On the contrary, seroconverted kidney transplant recipients showed impaired neutralization against emerging variants of concern after standard two-dose vaccination [52]. Comparing kidney to liver transplant patients after vaccination showed that in liver transplant recipients, IgG levels against every S epitope tested increased significantly compared to the kidney transplant recipients. It seems that vaccination elicits a stronger antibody response in liver than in kidney transplant recipients, a phenomenon that cannot entirely explained by the different immunosuppression employed [53]. In people infected with HIV (PLWH) and receiving suppressive antiretroviral therapy binding circulating antibodies against RBD were measured one month following the first and second COVID-19 vaccine doses, and again 3 months following the second dose. It was shown that PLWH with well-controlled viral loads and CD4+ T-cell counts in a healthy range generally mounted strong initial humoral responses to dual COVID-19 vaccination [54]. Overall, cancer patients with COVID-19 have reduced survival. While most cancer patients, have an almost 100% rate of seroconversion after SARS-CoV-2 infection or vaccination, patients with hematological malignancies have lower or even minimal seroconversion rates. A study from Florida [55] revealed that in 515 cancer patients seropositivity after two vaccination doses was 90.3% but was significantly lower among patients with hematologic cancer (84.7%) vs. solid tumors (98.1%) and was lowest among patients with lymphoid cancer (70.0%). Importantly, patients receiving vaccination within 6 months after anti-CD20 monoclonal antibody treatment had a significantly lower seroconversion (6.3%) compared with those treated 6 to 24 months earlier (53.3%) or those who never received anti-CD20 treatment (94.2%). In another prospective observational study immunogenicity was assessed in 85 patients treated with immune checkpoint inhibitors (ICIs) for a broad range of solid tumors. Despite the relatively poor responses following the priming dose, the seroconversion rates significantly increased after the second dose; the administration of a third booster dose remarkably optimized antibody responses [56]. Similarly, in a cohort of patients with hematologic malignancies, 76.3% of patients developed humoral immunity, and the cellular response rate was 79%. Hypogammaglobulinemia, lymphopenia, active hematologic treatment, and anti-CD20 therapy during the previous 6 months were associated with an inferior humoral response. A significant dissociation between the humoral and cellular responses was observed in patients treated with anti-CD20 therapy; in these cases, the humoral response was 17.5%, whereas the cellular response was 71.1%. In these patients, B-cell aplasia was confirmed while T-cell counts were preserved [57]. Finally, in a cohort where patients had received bone marrow transplantation or CAR-T cells, significantly lower anti-S antibodies were noted to the Wuhan strain following 2doses of the BNT162b2 mRNA vaccine, with proportional lower cross-recognition against Beta, Delta, and Omicron S-RBD proteins. Both cohorts neutralized the wildtype WA1 and Delta variants but not the Omicron variant [58]. The titers of neutralizing antibodies were also determined in patients with Multiple Myeloma (MM) or Waldenström macroglobulinemia (WM) after vaccination. Patients with MM produce lower amounts of neutralizing antibodies against SARS-CoV-2 after full vaccination, even after two booster doses, especially those under treatment with anti-CD38 or anti-BCMA therapies [59], [60], [61], [62]. In MM, vaccine-mediated antibody production is affected by race, disease, vaccine, and treatment characteristics [63]. In WM, the data suggest that vaccination with either 2 doses of the BNT162b2 or 1 dose of the AZD1222 vaccine led to lower production of neutralizing antibodies in patients compared to controls. Moreover, active treatment with either rituximab or Bruton's tyrosine kinase inhibitors was proven to be an independent prognostic factor for suboptimal antibody response after vaccination, even after a booster vaccine dose [64,65]. In patients with myeloid malignancy, including 46 patients with acute myeloid leukemia (AML) and 23 patients with myelodysplastic syndrome (MDS), seroconversion rates were 94.7% and 100% respectively, with no significant difference from healthy controls. Nevertheless, patients with MDS showed a significantly lower antibody titer than that found in healthy controls or AML patients. This study demonstrates that patients with myeloid malignancies may be more responsive to vaccines than patients with lymphoid malignancies [66]. In patients with chronic lymphocytic leukemia or B cell non-Hodgkin lymphoma, and multiple myeloma the use of a third vaccine dose is supported by evidence, even though some of these patients will still demonstrate vaccine failure [67]. Finally, in another cohort, and despite the absence of humoral immune responses in fully vaccinated anti-CD20-treated patients with lymphoma, their CD8+ T-cell responses reach similar frequencies and magnitudes as controls [68]. In a consensus generated by members of the European Multiple Myeloma Network it was confirmed that a suboptimal anti-SARS-CoV-2 humoral immune response, means that a proportion of patients are likely unprotected. Factors associated with poor response are uncontrolled disease, immunosuppression, concomitant therapy, more lines of therapy, and CD38 antibody-directed and B-cell maturation antigen-directed therapy. These facts suggest that monitoring the immune response to vaccination in patients with multiple myeloma might provide guidance for the administration of additional doses of the same or another vaccine, or even treatment discontinuation [69]. Overall, clearly, the subtype of hematologic malignancy and B-cell depleting treatment may predict a poor immune response to vaccination. Recently, antiviral drugs and monoclonal antibodies for pre-exposure or post-exposure prophylaxis and for early treatment of COVID-19 have become available. These therapies should be offered to patients at high risk for severe COVID-19 and vaccine non-responders, including patients with hematologic malignancy [70]. Evidence suggests that patients with hematologic cancer and those who are receiving immunosuppressive treatments may need additional vaccination doses [55]. There is clearly a need to develop guidelines to direct vaccination schedules and protective measures in oncology patients, differentiating those with hematological malignancies and those in an immunocompromised state [71].
Public health decisions require surveillance testing to obtain accurate epidemiological data for COVID-19 pandemic. Surveillance testing may be random sampling of a population to determine incidence and prevalence. To this end, testing need to be able to discriminate immunity from active infection versus from vaccination. In a population level, determining true rates of infection can inform us on the effectiveness of measures used for the restrain of the pandemic i.e., vaccination and social distancing. Similarly, such serological surveys can be performed in places of importance such as hospitals, nursing homes, critical workplaces and universities [72]. Regarding methods, these studies (unless otherwise stated) employ anti-S or anti-RBD determination with FDA-approved tests.
In a population level, interesting data emerged from Australia. As of mid-2021, Australia's only nationwide COVID-19 epidemic occurred in the first 6 months of the pandemic. In Australia's largest national SARS-CoV-2 serosurvey from 11,317 specimens only 71 were positive for SARS-CoV-2-specific antibodies while no seropositive specimens had neutralizing antibodies, thus the study concluded that Australia's seroprevalence was extremely low (<0.5%) and highlighted the population's naivety to the virus and the urgency for vaccine protection [73]. In another national-wide study from Mexico, and from 9640 blood samples, seroprevalence was estimated by socioeconomic and demographic characteristics. The national seroprevalence was 24.9% being lower for adults 60 years and older. Higher seroprevalence was found among urban and metropolitan areas, low socioeconomic status, low education and workers. Among seropositive people, 67.3% were asymptomatic. These data suggested that social distancing, lockdown measures and vaccination programs need to consider that vulnerable groups are more exposed to the virus [74]. In order to estimate the prevalence of unidentified SARS-CoV-2 infection in the general population of Hong Kong, a prospective cross-sectional study was conducted after each major wave of the COVID-19 pandemic. The study enrolled 4198 participants. Only six participants were confirmed to be positive for anti-SARS-CoV-2 IgG; the adjusted prevalence of unidentified infection was 0.15%. Extrapolating these findings to the whole population, indicated that there were fewer than 1.9 unidentified infections for every recorded confirmed case and it was estimated that the overall prevalence of SARS-CoV-2 infection in Hong Kong before the roll out of vaccination was less than 0.45% [75]. In a Norwegian population-based cross-sectional study, a total of 110,000 people aged 16 years or older were randomly selected during November-December 2020 (before vaccine introduction) and were invited to complete a questionnaire and provide a dried blood spot sample. National weighted and adjusted seroprevalence was 0.9%. In this paradigm, seroprevalence was comparable to virologically detected cases [76]. In Greece, a serosurvey was conducted between March and December 2020. It was designed as a cross-sectional survey repeated at monthly intervals. Of 55,947 serum samples collected, 705 (1.26%) were found positive for antibodies, with higher seroprevalence (9.09%) observed in December 2020. Highly populated metropolitan areas were characterized with elevated seroprevalence levels as compared to the rest of the country [77]. Interestingly, surveillance data in high-income countries have reported more frequent SARS-CoV-2 diagnoses in ethnic minority groups. To further test this, the cumulative incidence of SARS-CoV-2 was estimated in six ethnic groups in Amsterdam, the Netherlands. Compared to Dutch-origin participants (15·9%), cumulative SARS-CoV-2 incidence was higher in participants of South-Asian Surinamese, African Surinamese, Turkish, Moroccan and Ghanaian background. Also, SARS-CoV-2 incidence was higher in the largest ethnic minority groups of Amsterdam, particularly during the second wave. SARS-CoV-2 antibody seroprevalence can also add crucial epidemiological information about population infection dynamics [78]. To assess the evolving SARS-CoV-2 seroprevalence related to the first national lockdown in Belgium, a nationwide seroprevalence study was performed, using 3000–4000 residual samples during seven periods. Seroprevalence increased from 1.8% to 5.3% over a period of 3 weeks during lockdown (start lockdown mid-March 2020). Thereafter, seroprevalence stabilized. This showed that during lockdown, an initially small but increasing fraction of the Belgian population showed serologically detectable signs of exposure to SARS-CoV-2, which did not further increase when confinement measures eased and full lockdown was lifted [79]. In Germany, a federal state-wide cross-sectional seroprevalence study named SaarCoPS, representative for the adult population was performed. Serum was collected from 2940 adults via stationary or mobile teams during the 1st pandemic wave steady state period. They estimated an adult infection rate of 1.02%, an underreporting rate of 2.68-fold and infection fatality rates of 2.09% in all adults including elderly individuals. These type of studies are important because they can provide a valuable baseline for evaluation of future pandemic dynamics and impact of public health measures on virus spread and human health [80]. Such studies have also been published among others from Cyprus [81] and Malawi [82]. These kind of studies can help determine how previous SARS-CoV-2 infection and the time since vaccination should be considered when planning booster doses and the design of COVID-19 vaccine strategies [83].
In October 2020 SARS-CoV-2 seroprevalence among hospital healthcare workers (HCW) of two Irish hospitals was 15 and 4. 1%, respectively. In a comparative study in the same HCW population 6 months later, measuring anti-nucleocapsid and anti-spike antibodies, seroprevalence increased to 21 and 13%, respectively; 26% of infections were previously undiagnosed. Breakthrough infection occurred in 23/4111(0.6%) of fully vaccinated participants; all had anti-S antibodies [84]. HCWs in COVID-19 patient care in Sweden have been infected with SARS-CoV-2 at a higher rate compared to blood donors. This Swedish study detected substantial variation between different IgG-assays and proposed that multiple serological targets should be used to verify past infection. Their data suggested that CD4+ T-cell reactivity was not a suitable measure of past infection and does not reliably indicate protection from infection in naive individuals [85]. HCW in Switzerland, with exposure to COVID-19 patients had only a slightly higher absolute risk of seropositivity compared to those without, suggesting that the use of PPE and other measures aiming at reducing nosocomial viral transmission were effective. This study demonstrated that household contact with known COVID-19 cases represented the highest risk of seropositivity [86]. Studies of similar design have been reported from multiple places around the globe. A study in Colombian hospital workers [87] can be cited where seroprevalence was higher than measurable acute infection prevalence thus providing a way of determining true infection rates. Also such studies have been used to determine which group of healthcare providers are more prone to infection showing that those in acute medical units and those working closely with COVID-19 patients were at highest risk of infection [88]. This was also true for a Belgian hospital where seroprevalence was higher among participants in contact with patients or with COVID-19 confirmed subjects or, to a lesser extent, among those handling respiratory specimens, as well as among participants reporting an immunodeficiency or a previous or active hematological malignancy [89]. Similar findings have been reported from a Greek tertiary hospital where clinicians in contact with patients, as expected were more exposed-infected [90]. In a Japanese hospital, aiming to understand the mode of nosocomial infection, 685 HCW were recruited prior to the vaccination with anti-COVID-19 vaccine. Positive rates of HCW's working in COVID-19 wards were significantly higher than those of HCW's working in non-COVID-19 wards. By subtracting the positive rates of PCR from that of IgG (RBD), the rate of overall silent infection were estimated to be 6.0% [91]. In a cross-sectional study from hospital staff in a University Hospital in Munich, Germany, overall seroprevalence of SARS-CoV-2-IgG in 4554 participants was 2.4%. Staff engaged in direct patient care, including those working in COVID-19 units, had a similar probability of being seropositive as non-patient-facing staff. Increased probability of infection was observed in staff reporting interactions with SARS-CoV-2‒infected coworkers or private contacts or exposure to COVID-19 patients without appropriate personal protective equipment [92]. SARS-CoV-2 anti-S-IgG level were also measured in 535 vaccinated healthcare workers from Israel with known previous infection status 6–8 months after the second dose and it was shown that when interpreted alongside vaccination timing, anti-S serological assays could confirm or exclude previous infections within the previous 3 months [93]. In another approach, infection rates were calculated in fixed cohorts by PCR and antibody testing of 1% of the local population and >90,000 app-based dataset. The study surveilled a catchment area of 300,000 inhabitants. Increased risk for seropositivity was detected in several high-exposure groups, especially nurses. As probably expected, contact to a COVID-19-affected person was the strongest risk factor, whereas public transportation, having children in school, or tourism did not affect infection rates [94]. Seroprevalence was assessed among health workers in five public hospitals located in different geographic regions of Ethiopia. A total of 1997 sera were collected. The overall seroprevalence was 39.6%. Of the 821 seropositive HCWs, 224 had a history of symptoms consistent with COVID-19 while 436 had no contact with COVID-19 cases as well as no history of COVID-19 like symptoms. These findings highlight the significant burden of asymptomatic infection in Ethiopian hospitals and may reflect the scale of transmission in the general population [95]. In another sensitive population, seroprevalence in children less than 6 years of age was tested in the canton of Fribourg. A total of 871 children, with a median age of 33 months were included; 412 (47%) were female. Overall, 180 (21%) of children were seropositive. The number of household members tested positive for SARS-CoV-2 (PCR test) was the main exposure risk, but the family size was not associated with an increased risk of infection. In young children, extra-familial care does not increase the risk of becoming SARS-CoV-2 seropositive, neither does the number of contacts present in extra-familial care [96]. This approach can be useful for sensitive populations such as the people experiencing homelessness (PEH). In a Danish study it was shown that the prevalence of SARS-CoV-2 antibodies was more than twice as high among PEH and associated shelter workers, compared to the background population. These results could be taken into consideration when deciding in which phase PEH should become eligible for a vaccine [97]. Another sensitive population is nursing homes. In a Belgian study, seroprevalence was determined among residents and staff randomly selected from 20 nursing homes geographically distributed in Flanders, Belgium. The seroprevalence in the 20 nursing homes varied between 0.0% and 45.0%. This study showed that nursing homes are more affected by SARS-CoV-2 than the general population. The noted large variation suggests that some risk factors for the spread among residents and staff may be related to the nursing home itself and is a sign that epidemiological data in specialized places must be interpreted with caution [98]. In another Danish study, citizens living in social housing areas of low socioeconomic status had a three times higher SARS-CoV-2 seroprevalence compared to the general Danish population. The seroprevalence was significantly higher in males and increased slightly with age. Living in multiple generations households or in households of more than four persons was a strong risk factor for being seropositive [99].
Another insightful clinical application of anti-SARS-CoV-2 antibody testing is in assessing neurological disease associated with COVID-19. To investigate the pathophysiological mechanism of encephalopathy and prolonged comatose or stuporous state in severally ill COVID-19 patients, antibodies were measured in the CSF. All eight patients assayed had anti-SARS-CoV-2 antibodies in their CSF, while 4/8 patients had high titers which were comparable to high serum values. This was suggestive of blood-brain barrier (BBB) disruption; which likely eased the entry of cytokines and inflammatory mediators into the CNS enhancing neuroinflammation and neurodegeneration [3]. In a similar study, COVID-19 antibody responses were measured in serum and CSF samples from 16 patients with neurological symptoms. IgG specific for S was found in 81% of patients in serum and in 56% of patients in CSF. Interestingly, levels of IgGs in both serum and CSF were associated with disease severity and all patients with elevated markers of CNS damage in CSF also had anti-SARS-CoV-2 antibodies in the CSF; further anti-SARS-CoV-2 CSF antibodies had the highest predictive value for neuronal damage versus all tested clinical variables and biomarkers [100]. In a cross-sectional study of CSF neuroinflammatory profiles from 18 COVID-19 subjects with neurological complications (stroke, encephalopathy, headache), pro-inflammatory cytokines (IL-6, TNFα, IL-12p70) and IL-10 were increased only in the CSF of stroke COVID-19 subjects; a similar increase was also observed in non-COVID-19 stroke subjects. Anti-SARS-CoV-2 antibodies were observed in the CSF of 77% of COVID-19 patients with severe disease despite no evidence of SARS-CoV-2 viral RNA and CSF-CRP was present in all subjects with critical stages of COVID-19 (7/18) but only in 1/82 controls [101]. In another study, blood and CSF samples from 11 critically ill COVID-19 patients presented with unexplained neurological symptoms including myoclonus, oculomotor disturbance, delirium, dystonia and epileptic seizures, were analyzed for anti-neuronal and anti-glial autoantibodies. All patients showed anti-neuronal autoantibodies in either serum or CSF and antigens included known intracellular and neuronal surface antigens, but also various specific undetermined epitopes. These antigens were found to localize in vessel endothelium, astrocytes and neuropil of basal ganglia, hippocampus or olfactory bulb. The notion that several COVID-19 triggered autoantibodies may lurk in the shadows due to potential molecular mimicry of SARS-CoV-2 proteins with human polypeptides is pending confirmation. Any type of autoantibody may explain certain aspects of multi-organ disease in COVID-19 [102]. When CSF-derived monoclonal antibodies were isolated from an individual with severe COVID-19 it was found that these monoclonal antibodies targeted both antiviral and anti-neural antigens, including one clone that reacted to both spike protein and neural tissue [103]. Notably, in a distinct cohort of 60 prospective patients with encephalopathy and severe COVID-19 no autoantibodies were detected. These neuro-COVID-19 patients presented elevated levels of the cytokines IL-18, IL-6, and IL-8 in both serum and CSF, while MCP1 was elevated only in CSF and IL-10, IL-1RA, IP-10, MIG and NfL were increased only in serum. The levels of 14-3-3 and NfL in CSF significantly correlated with the degree of neurologic disability in the daily activities at the following 18 months [104]. These data combined do not support direct infection of the CNS by SARS-CoV-2 or specific neuroinflammation in the pathogenesis of neurological complications in COVID-19 [105]. Thus, the role and possible neural cross-reactivity of CSF anti-SARS-CoV2 IgG antibodies remains ambiguous. Evidence from CSF profiling in COVID-19 with neurological symptoms mainly suggests BBB disruption in the absence of intrathecal inflammation, compatible with cerebrospinal endotheliopathy. In that context, persistent BBB dysfunction and elevated cytokine levels may contribute to both acute symptoms and long-COVID [106]; therefore, measurement of anti-SARS-CoV-2 circulating IgGs in the CSF likely provides a reliable biomarker for the appearance of long-COVID symptoms.
As the pandemic continues, with new strains emerging, anti- SARS-CoV-2 antibody measuring will remain an indispensable tool. As discussed, through careful and on-point testing valuable information can be extracted on disease prognosis, prevention, epidemiology and care for sensitive individuals (immunocompromised) or sensitive populations (the elderly or hospitalized). In addition, as variants of concern will most likely keep emerging, determining different serotypes, as defined by the humoral immune response adds tools in the continuing global effort. The paradigm (and representing a significant evolutionary leap) of Omicron strain and its subvariants [107] teaches that it has antigenic features that clearly distinguish it from previous SARS-CoV-2 variants; therefore, some antibody tests are less sensitive against Omicron [108]. Updating tests as necessary, will ensure that we will maintain the ability to monitor SARS-CoV-2 seroprevalence in the community post-infection and/or vaccination [109], while in individual patients monitoring humoral immune responses aids disease prognosis. Worth mentioning is however, that given the acquisition of memory (B- and T-) immune cells post-infection and/or vaccination [110], the titers of circulating antibodies cannot entirely predict the protection levels of an individual from reinfection with an existing or a new strain. Moreover, given that tissues (including the mucosa) and not blood are the main sites of mounted immune responses upon microbial/viral infection (4) the validity of circulating antibodies in predicting overall immunity and protection against future infections should not be overestimated. |
||
PMC9647047 | Xuanzong Li,Ruozheng Wang,Shijiang Wang,Linlin Wang,Jinming Yu | Construction of a B cell-related gene pairs signature for predicting prognosis and immunotherapeutic response in non-small cell lung cancer | 27-10-2022 | non-small cell lung cancer,B cell marker genes,prognostic signature,immunotherapy,gene repair | Background Accumulating evidence indicates that the B cells play important roles in anti-tumor immunity and shaping tumor development. This study aimed to explore the expression profiles of B cell marker genes and construct a B cell-related gene pairs (BRGPs) signature associated with the prognosis and immunotherapeutic efficiency in non-small cell lung cancer (NSCLC) patients. Methods B cell-related marker genes in NSCLC were identified using single-cell RNA sequencing data. TCGA and GEO datasets were utilized to identify the prognostic BRGPs based on a novel algorithm of cyclically single pairing along with a 0-or-1 matrix. BRGPs signature was then constructed using Lasso-Cox regression model. Its prognostic value, associated immunogenomic features, putative molecular mechanism and predictive ability to immunotherapy were investigated in NSCLC patients. Results The BRGPs signature was composed of 23 BRGPs including 28 distinct B cell-related genes. This predictive signature demonstrated remarkable power in distinguishing good or poor prognosis and can serve as an independent prognostic factor for NSCLC patients in both training and validation cohorts. Furthermore, BRGPs signature was significantly associated with immune scores, tumor purity, clinicopathological characteristics and various tumor-infiltrating immune cells. Besides, we demonstrated that the tumor mutational burden scores and TIDE scores were positively correlated with the risk score of the model implying immune checkpoint blockade therapy may be more effective in NSCLC patients with high-risk scores. Conclusions This novel BRGPs signature can be used to assess the prognosis of NSCLC patients and may be useful in guiding immune checkpoint inhibitor treatment in our clinical practice. | Construction of a B cell-related gene pairs signature for predicting prognosis and immunotherapeutic response in non-small cell lung cancer
Accumulating evidence indicates that the B cells play important roles in anti-tumor immunity and shaping tumor development. This study aimed to explore the expression profiles of B cell marker genes and construct a B cell-related gene pairs (BRGPs) signature associated with the prognosis and immunotherapeutic efficiency in non-small cell lung cancer (NSCLC) patients.
B cell-related marker genes in NSCLC were identified using single-cell RNA sequencing data. TCGA and GEO datasets were utilized to identify the prognostic BRGPs based on a novel algorithm of cyclically single pairing along with a 0-or-1 matrix. BRGPs signature was then constructed using Lasso-Cox regression model. Its prognostic value, associated immunogenomic features, putative molecular mechanism and predictive ability to immunotherapy were investigated in NSCLC patients.
The BRGPs signature was composed of 23 BRGPs including 28 distinct B cell-related genes. This predictive signature demonstrated remarkable power in distinguishing good or poor prognosis and can serve as an independent prognostic factor for NSCLC patients in both training and validation cohorts. Furthermore, BRGPs signature was significantly associated with immune scores, tumor purity, clinicopathological characteristics and various tumor-infiltrating immune cells. Besides, we demonstrated that the tumor mutational burden scores and TIDE scores were positively correlated with the risk score of the model implying immune checkpoint blockade therapy may be more effective in NSCLC patients with high-risk scores.
This novel BRGPs signature can be used to assess the prognosis of NSCLC patients and may be useful in guiding immune checkpoint inhibitor treatment in our clinical practice.
Lung cancer is one of the most common cancers in the world, with a high mortality rate (1). Non-small cell lung cancer (NSCLC) accounts for 80-85% of all lung cancers and mainly consists of lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) subtypes (2). Even though innovative treatment strategies, including immunotherapy and molecular targeted therapy, have revolutionized the management model of NSCLC patients, the 5-year overall survival (OS) rate for this population remains less than 20% (3). The reliable and clinically applied biomarkers for prognosis in NSCLC patients with all histological subtypes are still very rare. In addition, there are no well-established predictive biomarkers for immunotherapy response until now. Therefore, identifying reliable biomarkers to predict survival and guide appropriate personalized treatment for NSCLC patients is necessary and urgent in our clinical practice. Accumulating evidence has revealed that tumor microenvironment (TME), accompanied by diverse tumor infiltrating lymphocytes (TILs), has been proven to play important roles in oncogenesis, tumor development and therapeutic efficacy prediction (4). In contrast to the well-investigated T cells (5), the potential role of tumor infiltrating B cells (TIL-B) is relatively less illustrated. Gottlin et al. found that the proliferative TIL-B could be identified in 35% of NSCLC, with significant variations in frequency across different clinical stages (6). B cells are a diverse population with highly heterogeneous subsets and functions (7). On the one hand, B cells can contribute to anti-tumor immunity by presenting antigens, producing antibodies, activating the complement cascade, assisting T-cell immune response, etc (8–10). On the other hand, there is also exist the regular B cells (Bregs) subset which can produce immunosuppression cytokines, such as IL-10 and IL-13, passively affects anti-tumor immunity (11). Recently, chen et al. demonstrated that the TIL-B has two major subtypes, namely the naïve-like and plasma-like B cells, with diverse functions in the progression of NSCLC (12). B cells were often associated with improved prognosis of NSCLC; however, the prognostic value of B cells is still controversial, with conflicting results across studies (13). Furthermore, several recent studies have found that B cells, associated mature tertiary lymphoid structures (TLSs) and plasma cells correlate with the efficacy of ICIs in multiple cancer types (14–19). Likewise, TLSs in tumors display substantial heterogeneity, and the prognostic and predictive value of TLSs is still controversial (20). However, a previous study demonstrated that only mature TLS with an active germinal center could predict the efficacy of immunotherapy in multiple cancer types (15). In fact, B cells are scarce in tumors without mature TLSs, whereas B cells are selectively activated and amplified in tumors with mature TLSs (21). Considering the significant roles of B cells in shaping the tumor immune environment and ICIs responses, therefore, it is necessary to make a comprehensive analysis of the heterogeneity, prognostic and immunotherapeutic predictive values of B cells in NSCLC. Single-cell RNA-sequencing (scRNA-seq) method provides a potent approach for us to explore the complex biological behavior of TILs and potential mechanisms for them in shaping tumor development in various cancer types (22–24). Hence, establishing B cell-related signatures by means of scRNA-seq data could be a useful way to predict immunotherapeutic responses and prognosis in NSCLC patients. In this study, we successfully constructed a B cells-related gene pairs (BRGPs) prognostic signature in NSCLC utilizing the gene pair approach and data from scRNA-seq and bulk RNA-sequencing public datasets. Importantly, this novel BRGPs signature with no dependence upon specific gene expression levels can improve risk stratification, prognosis accuracy and individualized immunotherapy for NSCLC patients.
We downloaded the transcriptome sequencing data and corresponding clinical features of NSCLC (LUAD and LUSC) patients from TCGA website (https://portal.gdc.cancer.gov/) on September 2021. A total of 1016 cases with tumor or normal sequencing data were included in this cohort. We only selected tumor sequencing data to construct a gene signature. The merged TCGA-NSCLC dataset was regarded as the training cohort. Then, we employed three microarray datasets (GSE37745, GSE30219 and GSE31210) from GEO database and set it as the validation cohort (25). Finally, a total of 999 and 570 NSCLC patients harboring both available gene expression and corresponding clinical data were included in the training and validation cohorts, respectively. The flowchart of the present study design is shown in Figure 1 .
A total of 22 cell clusters (C1-C22) and corresponding cluster-specific marker genes were retrieved from the additional files of one previous publication (12). Among these 22 cell clusters, C4 and C6 clusters were annotated as B cells, and their specific marker genes were utilized to be served as B cell-related genes (BRGs) ( Supplementary Table 1 ). In detail, a total of 90 unique genes including 35 marker genes from C4 (naïve-like B cells) subset and 59 marker genes from C6 (plasma-like B cells) subset were defined as BRGs in this study.
BRGs were screened out using a median absolute deviation (MAD) >0.5, as those genes showed high variation in the samples from entire training cohort. Of note, these BRGs were also available in the validation cohort. Next, we used the gene expression levels of these BRGs in each sample for a pairwise comparison to construct BRGPs. For one BRGP (gene A|gene B), if the expression value of gene A was greater than gene B, the score of this pair was considered as 1. Otherwise, the score of gene A|gene B was defined as 0. The score of each BRGP in all samples were calculated, and those BRGPs with 1 or 0 less than 20% or more than 80% of total samples were excluded, since these pairs had low variation.
Using “survival” and “survminer” R packages, we performed univariate Cox regression analysis to identify prognostic BRGPs with the limitation condition for P value less than 0.05 in the training cohort. Subsequently, using “glmnet” R package, the least absolute shrinkage and selection operator (LASSO) regression analysis was conducted to reduce the number of BRGPs and avoid model overfitting. Finally, the multivariate Cox regression analysis was performed to calculate the coefficients of the remaining BRGPs and construct prognostic signature. The risk scores of BRGPs signature for each NSCLC patient were calculated based on the value of these BRGPs (0 or 1) in the signature and weighted by multivariate Cox regression coefficient. The formula was as follows: risk score = ∑βi ×(BRG A|BRG B)i, where β is the regression coefficient.
Using “survivalROC” R package, the 1-, 2-, and 3-year receiver operating characteristic (ROC) curve analyses were performed, and corresponding values of the area under the curve (AUC) were also calculated. The point of maximum Youden Index in the 3-year ROC curve was defined as the optimal cut-off point of the risk score (26, 27). The formula was as follows: YoudenIndex = Sensitivity+ Specificity-1. Based on the optimal cut-off value of BRGPs, NSCLC patients in training and validation cohorts were classified into high- and low-risk groups, respectively. The Kaplan–Meier method and log-rank test were applied to compare the survival curves of different risk groups. Then, the prognostic value of the risk score as well as other characteristics, including age, gender, histology and stage, were evaluated by univariate and multivariate Cox regression analysis. Furthermore, the association between BRGPs signature and these characteristics was analyzed by chi-square test, and the result was displayed by heatmap. Besides, the differences of the distribution of the risk scores in NSCLC patients with different TNM stages were also compared by Wilcoxon rank-sum test.
Estimation of Stromal and Immune cells in Malignant Tumor tissues using Expression data (ESTIMATE) algorithm was employed to infer ESTIMATE, immune, and stromal scores and tumor purity based on “estimate” R package and gene transcriptional profiles (28). The distribution of the tumor purity, ESTIMATE, immune, and stromal scores were analyzed between high- and low- risk groups in NSCLC, respectively. Pearson correlation coefficient was used to compare the correlation relationships between the markers mentioned above and the risk score of BRGPs signature. CIBERSORT algorithm as well as the LM22 gene signature were used to calculate the abundance of 22 different immune cell types in each tumor sample (29). Sample deconvolution was performed 1000 permutations and P < 0.05 was required. Wilcoxon rank-sum test was used to compare the proportions of each tumor infiltrate immune cell subsets between high- and low- risk groups. Then, the Kaplan–Meier method and log-rank test were used to evaluate the prognostic values of different tumor infiltrate immune cell subsets in NSCLC patients. In addition, xCell and MCP-counter algorithms were also used to calculate the abundance of different immune cell types in high- and low- risk groups (30, 31).
Gene set enrichment analysis (GSEA) was performed to functionally elucidate the biological roles of the BRGPs in NSCLC. Using the Gene Ontology (GO) gene set (c5.all.v7.4.symbols.gmt) and Kyoto Encyclopedia of Genes and Genomes (KEGG) gene set (c2.cp.kegg.v7.4.symbols.gmt) from the Molecular Signatures Database, we analyzed the signaling pathway enrichment status in NSCLC patients with high- and low-risk scores by GSEA. To achieve a normalized enrichment score for each analysis, gene set permutations with 1,000 times were carried out. A nominal P < 0.05 and false discovery rate (FDR) < 0.05 were regarded as significant results. Furthermore, we compared the enrichment levels of 29 immune-related functional signatures between high- and low-risk groups based on the single sample gene set enrichment analysis (ssGSEA) algorithm in the GSVA R package (32, 33).
The association between PD-L1 mRNA (CD274) expression and the risk scores was evaluated by Wilcoxon test and spearman correlation analysis. The gene mutation data of LUAD and LUSC patients were downloaded from TCGA database (https://portal.gdc.cancer.gov/), and the tumor mutational burden (TMB) scores of each NSCLC patient were calculated as mutations per million bases. Then, the distribution of TMB in high- and low-risk groups was compared by Wilcoxon test, and spearman correlation analysis were performed between the risk score and TMB. Moreover, the somatic mutation features of high- and low-risk groups were visualized in the waterfall plot by “maftool” R package in LUAD and LUSC patients, respectively. Tumor Immune Dysfunction and Exclusion (TIDE) algorithm has been proven to have robust power for predicting clinical responses of ICIs treatment in melanoma, NSCLC and other cancer patients (34). Using the TIDE web (http://tide.dfci.harvard.edu), we obtained TIDE score, T cell dysfunction score and T cell exclusion score, and the distribution of those scores in high- and low-risk groups were compared by Wilcoxon test, respectively.
R software (version 4.1.1) was used to make all statistical analyses in this study, and P < 0.05 was considered statistically significant.
A total of 999 patients from the TCGA-NSCLC dataset were defined as training cohort in this study. Besides, three independent NSCLC cohorts from GEO database were analyzed as the validation cohorts. The characteristics of the NSCLC patients in training and validation cohorts were provided in Table 1 . Overall, in the training cohort, most patients over aged 65 years old (55.7%), were male (60.1%), had a disease stage I (54.4%), stage T2 (55.8%), stage N0 (64.2%), stage M0 (74.3%) and LUAD subtype (50.5%). Most patients with NSCLC in the GSE37745 cohort aged less than 65 years old (52.0%), were male (54.6%), in disease stage I (66.3%) and were LUAD (54.1%).
Ninety unique B cell-related marker genes were included in this study, and 327 BRGPs with substantial variation was eventually identified using the method of cyclically single pairing along with a 0-or-1 matrix. In the training cohort, a total of 47 BRGPs had significant prognostic values. Lasso-penalized multivariate Cox proportional hazards modeling was performed on these prognostic BRGPs to improve stability and accuracy. After 1000 iterations, we successfully established a 23 BRGPs signature, consisting of 28 unique BRGs ( Figures 2A, B ). The detailed information of the 23 BRGPs signature was shown in Table 2 . Besides, the expression levels of 28 unique BRGs were compared between NSCLC tumor and normal tissue in TCGA cohort, respectively. Importantly, most of BRGs (25/28) in BRGPs signature, except CCR7, HERUD1 and SEC11C, were differently expressed among NSCLC tumor and normal tissue ( Supplementary Figure 1 ). We then calculated the risk score of BRGPs signature for each NSCLC patient in the training and validation cohorts. The 1-, 2-, and 3-year ROC curves were generated to assess the accuracy of BRGPs signature in predicting the prognosis of NSCLC patients. And the results revealed that this model was efficient in predicting the prognosis of NSCLC patients as AUC values were all around 0.700 ( Figure 2C ). In addition, the time-dependent ROC curve was applied to determine the optimal cut-off value for dividing patients into high- and low-risk subgroups ( Figure 2D ). Furthermore, for NSCLC patients in the training cohort, the risk score histogram, survival status distribution, and each corresponding BRGPs value were plotted ( Figure 2E ). Our results indicated that the BRGPs signature could efficiently distinguish good or poor survival of patients with NSCLC (P < 0.001) in the training cohort ( Figure 2F ). Importantly, univariate and multivariate Cox regression analyses demonstrated that the risk score of BRGPs signature was significantly associated with poor prognosis and could serve as an independent prognostic factor in NSCLC patients ( Figure 3A, B ). Furthermore, we verified the prognostic value of our prediction signature in the validation cohort. As expected, the results showed that low-risk patients had a significant longer OS compared with high-risk patients, either in GSE37745 (P = 0.001), GSE30219 (P = 0.027) and GSE31210 (P = 0.031) ( Figure 2G and Supplementary Figure 2 ). And, the risk score was an independent prognostic factor based on univariate and multivariate Cox regression analyses in GSE37745 ( Figures 3C, D ).
The correlation between the BRGPs signature and clinicopathological characteristics of NSCLC patients was estimated using a chi-square test. Gender (P < 0.001), disease stage (P < 0.001), T stage (P < 0.001) and N stage (P < 0.05) were found to be significantly related to BRGPs signature ( Figure 4A ). Furthermore, we used Wilcoxon rank-sum test and demonstrated that NSCLC patients with stage III-IV, stage T3-4, stage N2-3, and stage M1 had significantly higher risk scores than patients in stage I-II (P < 0.001), stage T1-2 (P < 0.001), stage N0-1 (P = 0.023), and stage M0 (P = 0.0037) ( Figures 4B–E ).
Using ESTIMATE algorithm, we evaluated the differences in immunologic landscapes between high- and low-risk NSCLC patients. The results showed that ESTIMATE score, immune score and stromal score were significantly higher in low-risk NSCLC patients compared with their counterparts (all P < 0.001) ( Figures 5A–C ). By contrast, the tumor purity was significantly higher in the high-risk group (P < 0.001) ( Figure 5D ). Correspondingly, our findings suggested that ESTIMATE score, immune score and stromal score were all negatively correlated with the risk score (all P < 0.001) ( Figures 5E–G ), whereas the tumor purity was positively correlated with the risk score (P < 0.001) ( Figure 5H ). Using CIBERSORT method and LM22 single-cell gene expression model matrix, we compared the infiltration levels of 22 immune cells between high- and low-risk groups and the prognostic value of these immune cells in NSCLC patients. The relative expression landscape of these 22 immune cell types was described in each NSCLC patients ( Figure 6A ). We found that 12 immune cells were distributed with significant differences between the high- and low-risk groups ( Figure 6B ). Among these immune cells, the infiltration levels of CD8+ T cells, resting mast cells, plasma cells, resting dendritic cells, memory B cells, regulatory T cells (Tregs) and gamma delta T cells were higher in the low-risk group. Additionally, our findings indicated that CD8+ T cells, resting mast cells, plasma cells, resting dendritic cells and Tregs were all significantly associated with a favorable OS in patients with NSCLC ( Figures 6C–G ). On the contrary, the infiltration levels of neutrophils, resting NK cells, activated mast cells, M2 macrophages and M0 macrophages were higher in the high-risk group, and both were significantly associated with poor clinical outcomes in NSCLC patients ( Supplementary Figures 3A–E ). Furthermore, using additional immune deconvolution tools, we also demonstrated that several immune cells which primarily responsible for effective anti-tumor immunity, such as CD8+ T cell, CD4+ T cell and B-cells, were infiltrated higher in the low-risk group ( Supplementary Figures 4, 5 ).
To identify the underlying biological characteristics on the basis of BRGPs signature, we performed GSEA to predict the most significant enrichment signaling pathways between high- and low-risk NSCLC patients. Our results suggested that patients with low-risk scores significantly enriched with several immune activation related pathways, including activation of immune response, adaptive immune response based on somatic recombination of immune receptors built from immunoglobulin superfamily domains, antigen receptor mediated signaling pathway, B cell activation and B cell mediated immunity ( Figure 7A ). Meanwhile, the pathways involved in cell proliferation, such as nuclear chromosome segregation, sister chromatid segregation, mitotic sister chromatid segregation and helicase activity, were maximum extent enriched in the high-risk group ( Figure 7B ). Likewise, KEGG analysis found that several immune activation related pathways, such as intestinal immune network for IgA production and B cell receptor signaling pathway were enriched in low-risk BRGPs subgroup ( Supplementary Figure 6 ). We assessed the expression profiles of 29 immune-associated features to determine their immune-related signaling pathways, cell types, and functional activities. We quantified the level of enrichment of 29 immune signatures in each NSCLC sample using ssGSEA method. We demonstrated that the immune-associated biological behavior of patients in high- and low-risk groups was significantly different. Notably, patients in the low-risk group scored significantly higher in most immune or inflammation-related pathways, except for APC_co_inhibition, macrophages, MHC_class_I, NK_cells, parainflammation and Type_I_IFN_Reponse ( Figure 7C ).
Numerous studies indicated that patients with high PD-L1 expression and TMB scores have a higher chance of benefiting from ICIs treatment (35–39). As a result, we evaluated the association between BRGPs signature and these two well-characterized immunotherapy biomarkers. Unfortunately, there was no significant difference in PD-L1 mRNA expression between high- and low-risk NSCLC patients ( Figure 8A ). Besides, our results indicated that the risk score of BRGPs signature was not correlated with PD-L1 mRNA expression levels in TCGA-NSCLC, GSE30219 and GSE37745, but was positively correlated with PD-L1 mRNA expression in GSE31210 ( Figure 8B and Supplementary Figure 7 ). However, our findings indicated that NSCLC patients with high-risk score had significantly higher TMB scores (P < 0.001) ( Figure 8C ). Additionally, correlation analysis revealed a positive correlation between the risk score and TMB scores (P < 0.001) ( Figure 8D ). Moreover, the significant association between TMB score and the risk score of BRGPs signature was still existing in patients with either LUAD or LUSC (all P < 0.05) ( Supplementary Figures 8A, B ). Furthermore, the top 20 mutation genes of the high- and low-risk cohorts of LUAD and LUSC patients were plotted ( Figures 8E–H ). Then, we evaluated the relationship between BRGPs risk signature and TIDE-related scores. Interestingly, patients with high-risk scores had significantly higher exclusion scores, lower TIDE scores and lower T cell dysfunction scores compared with low-risk patients (all P < 0.01) ( Figures 8I–K ), implying that high-risk NSCLC patients may be more sensitive to immunotherapy. Unsurprisingly, an inferior survival rate for low-risk patients after immunotherapy were observed in GSE135222 ( Supplementary Figure 9 ). Collectively, these findings indicate that patients with high-risk scores are more likely to benefit from immunotherapy and that BRGPs may serve as a potential biomarker for predicting immunotherapy efficacy in NSCLC patients.
The tremendous clinical success of cancer immunotherapy refocused attention on various TILs, however, reliable biomarkers based on the TILs to predict immunotherapy response and prognosis of NSCLC patients are still very rare (4). In this study, we obtained B cell specific marker genes from a scRNA-seq study and innovatively conducted a new method of cyclically single pairing along with a 0-or-1 matrix to construct a novel BRGPs signature in NSCLC patients. In the training and validation cohorts, our novel BRGPs signature demonstrated effective prognostic performance and can be used as an independent risk factor for NSCLC patients. Analysis of clinicopathological characteristics, TME conditions, immune profiles and biological pathway revealed that patients with a low-risk score were characterized by early clinical stage, low tumor purity, high anti-tumor immune cell infiltration and immune-active states. Additionally, we found that patients with high-risk scores had significantly higher TMB scores and lower TIDE scores compared with patients with low-risk scores, which indicates that high-risk patients are more likely to benefit from immunotherapy. Collectively, BRGPs signature might be a useful biomarker to predict prognosis and immunotherapeutic effect in NSCLC patients. More importantly, our novel BRGPs signature only needs to detect the higher or lower expression level of the two BRG in each BRGP without requiring quantitative gene expression profiles (40), which avoids potential technical bias and improves its clinical practicability. In this study, the BRGPs signature was composed of 23 BRGPs, including 28 different BRGs. In the signature model, gene pairs (BIRC3|RGS16 and HES1|ITM2C) harbored the highest coefficients and presented positive and negative effects on the prognosis of NSCLC patients, respectively. BIRC3 acts as a member of inhibitors of apoptosis proteins (IAPs) family and plays an important role in pro-survival and antiapoptotic on the cells, which has been characterized in multiple cancer types (41). In LUAD, increased expression of BIRC3 could promote tumor growth and metastasis (42). RGS16 is one of the regulators of G protein singling (RGS) gene family members and negatively regulates G protein–coupled receptor (GPCR) signaling cascades (43). It was reported that RGS16 played central roles in immune and inflammatory responses (44, 45). Importantly, RGS16 can inhibit the Ras-Raf-MEK-Erk signaling cascade and promotes antitumor CD8+ T cell exhaustion (46). HES1, a Notch signaling pathway target, plays both oncogenic and tumor suppressor roles in different cell types (47). Interestingly, HES1, associated with Notch activation, was essential to inhibit the progression of B-cell acute lymphoblastic leukemia rather than T-cell acute lymphoblastic leukemia (47). Besides, HES1 has been shown to be positively correlated with the expression of FOXP3 and plays an important role in regulating the invasive and migratory functions of FOXP3 in NSCLC cells (48). ITM2C belongs to the Type II Integral Membrane protein (ITM2) family and is thought to be negatively regulates the amyloid-beta peptide production (49, 50). Importantly, ITM2C is highly and selectively expressed by Antibody Secreting Cells in the immune system (50). The signature genes identified in this study can provide potential targets for experimental design to give new insights into the pathological mechanisms in NSCLC. We performed 1-, 2-, and 3-year ROC curves analysis to assess the efficacy and accuracy of the BRGPs signature, and the corresponding AUC values were all close to 0.700, indicating that our predictive signature was effective in predicting the prognosis of NSCLC patients. Zhang et al. identified a 13-gene B cell-associated signature in LUAD patients, with 2-year AUC of 0.621 in the training cohort, inferior to the AUCs in our study (51). Additionally, a previous study demonstrated significant differences in the expression levels of B cell-related genes between patients with LUSC who had a good survival outcome and those who had a poor survival outcome (52). In this study, the risk score of our BRGPs signature was an independent prognostic factor in NSCLC patients, and we found that the risk signature was significantly associated with the clinical stage of NSCLC patients. These findings revealed that the major clinical significance of the BRGPs signature and prompted us to explore the potential underlying mechanism. Considering the remarkable impact of TME on the prognosis of cancer patients (53), we investigated the discrepancy in immune cell infiltration between low- and high-risk NSCLC patients. Notably, we found significant TME heterogeneity between high- and low-risk NSCLC patients using ESTIMATE and CIBERSORT methods. For example, our findings indicated that low-risk NSCLC patients had a higher proportion of CD8+ T cells, but a lower proportion of M2 macrophages. CD8+T cells have been linked to a better prognosis of patients with multiple cancer types (54). As the key effectors in the anti-tumor process, CD8+T cells can release perforin and granzyme and mediate cytotoxicity via Fas/FasL signaling pathway (55). Otherwise, the macrophages can be classified into M1 and M2 subtypes based on differentiation status and functional roles (54). Tumor-associated macrophages (TAMs) typically exhibit an M2-like phenotype which can secrete various immune suppress factors, including IL-10, TGFβ, and proangiogenic factors, and previous research has established a link between TAMs and disease progression and poor prognosis of NSCLC patients (56, 57). Then, the functional enrichment analysis revealed that immune-activating pathways were significantly enriched in low-risk NSCLC patients, whereas high-risk NSCLC patients were closely implicated in cell proliferation related functions. Indeed, several important BRGs found in BRGPs signature, such as BIRC3, IFT57, GADD45B and SPAG4, have been associated with the proliferation or migration of NSCLC cells (41, 42, 58–61). Therefore, high-risk NSCLC patients are more likely to harbor genome instability status and associated with high TMB, tumor progression, and relative advanced tumor stage. Collectively, BRGPs signature showed significant prognostic value in patients with NSCLC, and the potential biological mechanism may attribute to the dysregulation of the cell cycle and TME heterogeneity. Currently, immunotherapy, especially for immune checkpoint blockade, has revolutionized the treatment of lung cancer (62). However, the response rate of ICIs is relatively low, and most NSCLC patients cannot benefit from these immunotherapeutic agents (63). Therefore, developing reliable biomarkers to improve the prognosis of NSCLC with ICIs treatment is urgently needed. Up to now, various biomarkers have been investigated to determine the therapeutic effect of ICIs (64, 65). For instance, PD-L1 expression and TMB scores have been demonstrated to be independently associated with the efficacy of ICIs and can be used to guide ICIs treatment in our clinical practice. Likewise, TIDE methods are widely used for immunotherapeutic prediction and have been proven to have impressive predictive performance in various cancers (34, 66–68). The relationship between above mentioned ICIs-related biomarkers and BRGPs signature was investigated in this study. Our findings indicated that TMB scores rather than PD-L1 mRNA expression were positively correlated with the risk score. In contrast to the unfavorable prognosis associated with high TMB scores in NSCLC patients (69), TMB is common positively correlated with the improved efficacy of immunotherapy (70). Unfortunately, a positive correlation between BRGPs signature and PD-L1 mRNA expression was not found in all NSCLC cohorts. Since PD-L1 tumor staining by immunohistochemical is routinely used as an immunotherapy biomarker in multiple cancer types including NSCLC, further studies to investigate the relationship between BRGPs signature and PD-L1 expression are urgently warranted both in the mRNA and protein levels. Importantly, we found that NSCLC patients with high risk-scores had significantly higher TMB scores but lower TIDE scores, implying a greater potential for immunotherapy benefit. Hence, ICIs treatment may be a better option for NSCLC patients with high-risk scores. Nevertheless, the predictive value of BRGPs serving as a reliable biomarker in immunotherapy requires further validation. Undeniably, several limitations were existed in this study. Even though the prognostic value of our BRGPs signature was fully validated in TCGA and GEO cohorts, the study’ retrospective nature and the potential bias should not be neglected. Next, the results were achieved based on public database. Therefore, additional experimental studies (both in vitro and in vivo) are warranted to verify the molecular mechanism through which B cell-related genes affect NSCLC, and external clinical studies should be performed to further clarify the predictive capability of our BRGPs signature in NSCLC patients with and without immunotherapy. In conclusion, we established a novel BRGPs signature that could serve as a potent prognostic biomarker and a potential indicator of immunotherapeutic response in NSCLC. Importantly, our BRGPs signature significantly correlated with TME and TMB, indicating that these molecular changes might explain the clinical significance. Nonetheless, future clinical studies will be required to validate the utility of the constructed BRGPs signature as soon as possible.
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .
XL, SW and LW contributed to the study concept and design, and critical revision of the manuscript for important intellectual content. XL performed the data analysis and drafted the manuscript. All authors contributed to the article and approved the submitted version.
This work was partially supported by funds from the Natural Science Foundation of Shandong Province (ZR2019LZL012, ZR201911040452), the Academic Promotion Program of Shandong First Medical University (2019ZL002), Research Unit of Radiation Oncology, Chinese Academy of Medical Sciences (2019RU071) and the foundation of National Natural Science Foundation of China (82172865, 81627901, 81972863 and 82030082).
We would like to thank Freescience for English language editing.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647048 | Sean M. Morris,Abhishek Chauhan | The role of platelet mediated thromboinflammation in acute liver injury | 27-10-2022 | platelet,innate immunity,liver failure- therapy,acute on chronic liver failure (ACLF),ALF,liver,thromboinflammation,, immunothrombosis | Acute liver injuries have wide and varied etiologies and they occur both in patients with and without pre-existent chronic liver disease. Whilst the pathophysiological mechanisms remain distinct, both acute and acute-on-chronic liver injury is typified by deranged serum transaminase levels and if severe or persistent can result in liver failure manifest by a combination of jaundice, coagulopathy and encephalopathy. It is well established that platelets exhibit diverse functions as immune cells and are active participants in inflammation through processes including immunothrombosis or thromboinflammation. Growing evidence suggests platelets play a dualistic role in liver inflammation, shaping the immune response through direct interactions and release of soluble mediators modulating function of liver sinusoidal endothelial cells, stromal cells as well as migrating and tissue-resident leucocytes. Elucidating the pathways involved in initiation, propagation and resolution of the immune response are of interest to identify therapeutic targets. In this review the provocative role of platelets is outlined, highlighting beneficial and detrimental effects in a spatial, temporal and disease-specific manner. | The role of platelet mediated thromboinflammation in acute liver injury
Acute liver injuries have wide and varied etiologies and they occur both in patients with and without pre-existent chronic liver disease. Whilst the pathophysiological mechanisms remain distinct, both acute and acute-on-chronic liver injury is typified by deranged serum transaminase levels and if severe or persistent can result in liver failure manifest by a combination of jaundice, coagulopathy and encephalopathy. It is well established that platelets exhibit diverse functions as immune cells and are active participants in inflammation through processes including immunothrombosis or thromboinflammation. Growing evidence suggests platelets play a dualistic role in liver inflammation, shaping the immune response through direct interactions and release of soluble mediators modulating function of liver sinusoidal endothelial cells, stromal cells as well as migrating and tissue-resident leucocytes. Elucidating the pathways involved in initiation, propagation and resolution of the immune response are of interest to identify therapeutic targets. In this review the provocative role of platelets is outlined, highlighting beneficial and detrimental effects in a spatial, temporal and disease-specific manner.
Acute liver inflammation as a result of hepatic injury can occur both in patients with and without pre-existent liver disease. Acute failure (ALF) occurs in the former, specifically when a patient with no antecedent history of liver disease develops significant liver dysfunction due to an acute hepatocellular insult (1). ALF is characterized by deranged liver biochemistry, jaundice, coagulopathy and encephalopathy (1). Viral hepatitides are the leading cause of ALF in the Asia and Africa whilst in the West the majority of ALF due to drug-induced liver injury [acetaminophen (APAP) overdose] (1). Treatments options for APAP-induced ALF are limited (2) and treatment is therefore based largely around supportive care often in an intensive care setting; ultimately liver transplantation may be required (1). Ten percent of liver transplants in the West are due to acute liver failure (1). Dilemmas surrounding timing of liver transplantation, prognosis and donor shortage highlight the need for disease modifying treatments in patients with ALF (3). Acute-on-chronic liver failure (ACLF) on the other hand occurs in patients with established chronic liver disease or cirrhosis. Multiple stages of cirrhosis are recognized; in the compensated cirrhosis phase symptoms are modest and mortality low. Acute liver injuries result in hepatic and systemic inflammation in cirrhotic patients which then drive hepatic decompensation as manifest by the development of one or a combination of ascites, portal hypertensive gastrointestinal bleeding and or encephalopathy (4). The one year mortality increases from 3% in patients with compensated cirrhosis to 57% with acute decompensation; median patient survival falls from 12 years to only 2 years (4). There is a subgroup of patients that after acutely decompensating develop progressive extrahepatic organ failures on a background of severe systemic inflammation (5), this syndrome is referred to as ACLF (6) and has a short term mortality as high as 80% at 28 days (7); ACLF in fact represents the leading cause of mortality amongst patients with decompensated cirrhosis (7). Precipitating events for decompensation and eventually ACLF in patients with established cirrhosis include drugs, infections, bleeds and flares of the underlying disease (4). There are differences and similarities in the pathophysiological responses that underpin the development of ALF and ACLF, both involve a complex interplay between damaged hepatocytes, liver sinusoidal endothelial cells (LSECs), resident and circulating immune cells that initiate and potentiate inflammation but also determine resolution. Platelets are key protagonists in a number of these processes. Known historically for their hemostatic function at sites of vascular injury, it is now well established that platelets participate actively in the immune response intimately linking thrombosis and inflammation, in a process described as thromboinflammation (8–11). Whilst platelets are clearly involved in acute liver inflammation (12, 13), the involvement is likely to be stage and location specific varying in patients with and without pre-existent liver disease (14, 15). Of particular interest in the treatment of liver diseases is defining coagulopathy sparing platelet activation pathways (12, 16). Here we discuss the dualistic roles of platelets in the initiation, amplification and resolution of acute liver injury and how this drives the development of ALF and ACLF.
Sterile inflammation is a response to host cell damage in the absence of pathogens, a key step in restoration of homeostasis, however, becomes a pathological process in several causes of ALF including drug-induced liver injury, alcoholic hepatitis, non-alcoholic steatohepatitis and ischemia-reperfusion injury (17, 18). Central to this condition is the release and recognition of damage-associated molecular patterns (DAMPs) via pattern recognition receptors (PRRs). DAMPs are self-molecules with an ability to activate inflammation (19) including a number of different proteins, nucleic acids and mitochondrial components (20). They are released in inflammatory cell death, pyroptosis, necrosis and necroptosis (18, 21), in addition to non-inflammatory cell death during secondary necrosis of apoptotic bodies (22). PRRs are highly conserved receptors, originally discovered for their role in responding to pathogen associated molecular patterns (PAMPs), of which Toll-like receptors (TLRs) are best studied, responsible for promoting inflammation through cytokine production, chemokine production and expression of ligands involved in leucocyte adhesion and activation (17). In the liver, immune surveillance is performed by a number of resident and circulating leucocytes. Kupffer cells are particularly important sentinel macrophages of the liver participating in the immune response through detection of injury, leucocyte recruitment and mediate tissue repair (23). Importantly it has been demonstrated that platelets are the first cells to accumulate at sites of injury within the liver, thus generating interest into their role in the disease process (24).
In ACLF the main driver of widespread tissue injury is a systemic hyperinflammatory response (25), arising from massive release of inflammatory mediators including cytokines, chemokines, growth factors and bioactive lipid mediators. This leads to immune cell activation and subsequent immune-mediated tissue damage (26). Little is known about exact triggers however is likely to involve recognition of both PAMPs and DAMPs via PRRs. Approximately one-third of cases involve bacterial infections (27), attributed to bacterial translocation across the intestinal lumen (28). Inflammatory cell death, necroptosis and pyroptosis, is common in advanced liver disease triggering release of DAMPs propagating the immune response (29). The hyperinflammatory response often co-exists with innate immune dysfunction at humoral, physical and cell-mediated level (26). The condition is characterised by increased pro- and anti-inflammatory mediators (30), with the prevailing phenotype temporally and spatially dependent, although immunodeficiency has greater importance in advanced disease (31). Bernsmeier et al. (32), proposed a model whereby exaggerated inflammatory responses to DAMPs and PAMPs in cirrhosis promotes polarization of monocytes/macrophages to immunoregulatory phenotypes. In the presence of endothelial dysfunction, reverse migration of these regulatory cells leads to population expansion in distant organs and global immunosuppression. One might hypothesize that populations of exhausted leucocytes predispose patients to sepsis, driving expansion and activation of naïve innate immune cells potentiating cell damage. Dissecting these pathways, and elucidating the role of platelets, if any, is key to identify suitable therapeutic targets.
Platelets possess a range of receptors and a diverse proteasome facilitating interactions with endothelial cells, immune cells, the extracellular matrix and other platelets ( Figure 1 ) (9, 10, 33). Activation leads to platelet degranulation, through classical glycoprotein (GP) pathways at sites of vascular injury (9, 10) and more recently discovered alternative pathways (16), releasing cytokines, chemokines, vasoactive substances, growth factors and platelet-derived extracellular vesicles (PEVs) (containing microvesicles, exosomes and apoptotic bodies) (10, 34). Platelets exhibit dual roles in inflammation with pro- and anti-inflammatory effector functions, which are likely to be disease, organ and time-specific (10). They promote leucocyte recruitment and modulate effector functions through direct interactions (P-selectin-P-selectin glycoprotein ligand 1 (PSGL-1), GPIbα-macrophage-1 antigen (MAC-1, a complement receptor), GPIIbIIIa-MAC-1 through fibrinogen and CD40-CD40L pairings (35, 36)), chemokine/cytokine secretion and increasing vascular permeability (9, 10). They provide a link between innate and adaptive immunity, supporting antigen presentation and lymphocyte function (37). At sites of inflammation platelets also limit bleeding (38) through physical sealing and tightening of endothelial junctions (39). During injury resolution, platelets promote regeneration and homeostasis through chemokine, angiogenic factor and growth factor release (9). The interaction between platelets and leucocytes is bidirectional – leucocytes also promote platelet activation and, in turn, the coagulation cascade; this process is termed immunothrombosis (9). Initially recognized as a means to potentially limit pathogen spread and enhance clearance; there is now increasing recognition of the role microthrombi can play in liver pathobiology by inducing endothelial dysfunction and organ damage (13, 40). The role for thromboinflammation is evident in conditions traditionally associated with thrombosis including atherosclerosis (41), deep vein thrombosis (42) and reperfusion injury after ischemic stroke (43). There is also increasing evidence for a protective role in sepsis, limiting tissue injury and promoting pathogen clearance (37, 44–48). Clinically, thrombocytopenia is associated with poor outcomes in sepsis (44, 46) and platelet transfusion improves bacterial clearance through macrophage recruitment (48). With a wide array of PRRs platelets aid with pathogen clearance during viral infection, but also exacerbate tissue injury in response to DAMPs (49). Interestingly platelets demonstrate plasticity in function, as evidenced in major trauma, emphasizing temporally and spatially diverse roles in the immune response (50). In the acute phase of injury platelets are poorly responsive to activation ex vivo, contributing to coagulopathy, followed by hyper-responsiveness exhibiting a pro-inflammatory and prothrombotic phenotype resulting in secondary organ damage (50). Pathways modulating platelet immune functions without impairing normal hemostasis are attractive therapeutic targets. Whilst this is particularly relevant in acute liver injury where patients can have an unpredictable coagulopathy, the bleeding diathesis in ALF is arguably an overstated concern but beyond the scope of this current review (51, 52). Immunoreceptor tyrosine-based activation motif receptors C-type lectin-like receptor 2 (CLEC2) and GPVI share similar downstream pathways involved in platelet activation ( Figure 2 ) (16). CLEC2, activated by endogenous ligand podoplanin, is important in hemostasis although, promisingly, blockade does not produce a hemorrhagic phenotype (53). On the other hand, the podoplanin-CLEC2 axis appears to be vital in several models of thromboinflammation (16). In infection these receptors have both beneficial and detrimental roles (13, 45–47, 54). GPVI promotes neutrophil recruitment during pneumonia (54), whilst inflammatory macrophages promote platelet aggregation via CLEC2 leading to pathogenic thrombosis in peritoneal sepsis (13). In other mouse models targeting the podoplanin-CLEC2 axis may have a role in limiting immune activation (45–47). In the liver, platelets promote hemostasis, fine tune the immune response through direct and indirect interactions and serve as a reservoir of biologically active substances (15). Unsurprisingly, platelet effector functions are complex and dualistic – often disease and stage specific (14).
LSECs constitute a unique vascular bed with a powerful scavenger system and potent endocytic capacity, aptly placed to respond appropriately to an array of antigens balancing immune activation and tolerance (55). LSECs orchestrate the immune response but also interact with hepatocytes and hepatic stellate cells in the process of regeneration or fibrosis (55). The hepatic sinusoids are classically narrow, characterised by low shear stress and thus initial recruitment is often selectin independent. The sinusoids express minimal levels of selectins in vivo (56, 57). Platelet sequestration is observed within hepatic sinusoids in numerous models of inflammation including APAP-induced (12), non-alcoholic steatohepatitis (58), ischemia-reperfusion injury (59, 60) and viral hepatitis (61). Platelet-endothelial interactions are also bidirectional and vary between type of injury encountered. Due to the diverse range of receptors, surface ligands and ability to secrete soluble mediators, both cell types participate in recruitment, adhesion, and activation of immune cells ( Figure 3A ). In vitro studies have demonstrated platelet adhesion is, in part, integrin mediated (through GPIIbIIIa and αVβ3) (62). Platelet adhesion triggers CXCL8 and CCL2 production by endothelial cells, promoting leucocyte recruitment. In a model of liver ischemia-reperfusion injury, LSECs are highly susceptible during cold preservation (63). Depletion of adenosine triphosphate (ATP) impairs transmembrane active ion transport leading to cellular swelling and mitochondrial dysfunction (64). On reperfusion reactive oxygen species production depletes free radical scavengers, leading to activation with increased expression of P-selectin (60). Upregulation of P-selectin promotes platelet adhesion, activation and LSEC apoptosis (65). In a bile duct (BDL) model of cholestatic liver injury of mice, platelet accumulation promotes leucocyte sequestration and hepatocyte damage in a partially p-selectin dependent manner (66). In this model, the role of LSEC podoplanin expression has been explored (67). Podoplanin expression is increased in BDL-treated mice and those pre-treated with anti-CLEC2 antibodies had reduced hepatic inflammation and subsequent fibrosis. The authors hypothesized platelet-derived serotonin (5-HT) released on activation of CLEC2 was responsible for reduced injury and promoting regeneration. Plasma serotonin was found to be raised in untreated BDL-mice. PEVs have been shown to induce endothelial cell apoptosis in models of sepsis (68). In a study of patients presenting with ALF (50% APAP-induced), increased levels of circulating microparticles were associated with presence of systemic inflammatory response syndrome, high-grade hepatic encephalopathy and death or requirement for liver transplantation (69). This may implicate PEVs in driving inflammation in ALF or simply highlight their utility as a marker of systemic inflammation. In ischemia-reperfusion injury, PEV levels increase rapidly at initiation of injury (70) and inhibition of microparticle release has been shown to reduce degree of injury (71). Endothelial cells interact with resident and migrating leucocytes to promote platelet adhesion and activation. In a model of ischemia-reperfusion injury, migrating CD4 T cells were shown to interact with endothelial cells to promote platelet adhesion (72) and Kupffer cells produced a similar effect via tumor necrosis factor (TNF)-α secretion (73). Bidirectional communication promotes a self-perpetuating cycle of inflammation leading to significant immune-mediated damage. These murine and human data reveal how platelets modulate the inflammatory landscape in acute liver injury in non-fibrotic livers to potentially drive ALF.
Evidence for thromboinflammation mediated liver damage in acute liver injury has clearly been demonstrated in viral models of murine hepatitis. Activated platelets in these models reduce sinusoidal blood flow through the secretion of vasoactive mediators including 5-HT, which then drives the development of inflammatory intrahepatic microthrombosis. Thromboinflammation thus directly delays effector cell recruitment reducing viral clearance and enhancing liver damage (74). During homeostasis, surface expression of CD39 on LSECs cleave ATP and adenosine diphosphate to adenosine monophosphate limiting platelet activation (75). In reperfusion injury, for example during ischemic hepatitis, CD39 expression is reduced promoting platelet activation via ATP. Injury is again potentiated through increased vascular tone and reduced blood flow from platelet-derived thromboxane A2 and 5-HT (59). Historic models reveal that during acute liver injury in rats endothelial damage causes the deposition of platelet rich thrombi within hepatic sinusoids (76). These models demonstrate that thromboinflammation disrupts hepatic microcirculation (77); reduced sinusoidal blood flow exacerbates hepatocyte dysfunction, increases DAMP expression and amplifies inflammation. These data provide a cogent explanation to how platelet driven thromboinflammation can drive both de novo acute liver failure for instance in virus induced ALF but also decompensation and ACLF in patients who contract a viral infection or suffer an ischemic hepatitis on a background of established liver cirrhosis. Non-selective beta blockers are used in cirrhosis to reduce portal pressure. They also exert beneficial non-hemodynamic effects in relation to bacterial translocation, particularly reducing incidence of ACLF secondary to bacteria-induced systemic inflammation (CANONIC study) (78). Recently reduced von Willebrand Factor (vWF) levels, as a marker of endothelial dysfunction, were identified as a possible marker of non-hemodynamic non-selective beta blocker effect (79). This may illustrate a link between endothelial dysfunction, platelets and bacteria-induced systemic inflammation in ACLF. More recently, examination into predictors of ACLF development during acute decompensation identified severity of inflammation as a key determinant of the development of portal venous thrombosis; again highlighting the link between inflammation and thrombosis in the context of liver cirrhosis (80). Further evidence for the role of immunothrombosis is observed in severity of decompensated cirrhosis (81). Here, portal and hepatic vein sampling was performed in patients undergoing transjugular intrahepatic portal systemic shunt insertion and demonstrated higher markers of platelet activation, lipopolysaccharide levels and inducible nitric oxide synthetase in the portal vein. This correlated with clinical disease severity, linking bacterial translocation and platelet activation with progression of disease. The specific molecular endothelial triggers for platelet activation within the liver and how this varies when comparing cirrhotic to non-cirrhotic livers merits further investigation.
Neutrophils are one of the key effector cells in the innate immune system responsible for driving inflammation and tissue injury in the liver (82–86). They are essential to host defense (87) but can also contribute to indiscriminate tissue injury and are responsible for cell damage in many diseases (88, 89). Neutrophil recruitment and activation is evident in a multitude of acute liver injuries which can drive acute liver failure including drug-induced liver injury (82), and ischemia-reperfusion injury (86) but also ACLF including viral hepatitis (83), alcoholic hepatitis (84), and non-alcoholic steatohepatitis (85). Neutrophils infiltrate the site of liver injury within minutes to hours via stimulated LSECs involves a process of selectin-mediated rolling, integrin-mediated firm adhesion followed by transendothelial migration in a well-described recruitment cascade (89). Recruitment to the liver during inflammation is influenced by resident Kupffer cells (90–92), LSECs (93) and stromal cells (94). Necrotic cells release necrotaxis signals including mitochondrial formylated peptides to facilitate precise homing (91). Neutrophils migrate through intravascular channels and exhibit swarming behaviour at the site of injury (91). In addition to initiating and propagating inflammation, neutrophils promote resolution and a return to homeostasis in focal thermal (24), APAP- (12) and carbon tetrachloride-induced (95) liver injury through clearance of cellular debris, production of extracellular matrix and normal revascularization. During ACLF neutrophils exhibit an exhausted phenotype – characterised by high levels of activation (96) but impaired core functions including phagocytosis, reactive oxygen specifies production and degranulation (97). Lower CXCR1/2 expression on neutrophils, a key chemotactic receptor, predicted poor outcome in hepatitis B-virus (HBV) related ACLF (98). On the other hand, neutrophil-to-lymphocyte ratio has been shown to be an independent predictor of prognosis in HBV ACLF (99). Interestingly, patients with ratios ≥3 had lower mortality, but those with >6 were at greater risk of mortality in ACLF.
Platelets contribute to neutrophil recruitment and activation through ligand-receptor interactions and chemokine secretion ( Figure 3B ) (10, 100, 101). Initial interactions between platelet P-selectin and neutrophil PSGL-1 are critical for recruitment, activation of MAC-1 and LFA-1 (lymphocyte function-associated antigen 1, an integrin expressed on leucocytes) and release of neutrophil extracellular traps (NETs) (102, 103). The importance of platelet-neutrophil interactions via p-selectin is illustrated by improved survival of p-selectin deficient mice in an ischemia-reperfusion injury model of acute liver injury (104). MAC-1 allows direct binding with platelets via GPIbα and indirectly to GPIIbIIIa via fibrinogen (105). Recruitment is further amplified through secretion of cytokines, chemokines and growth factors including platelet factor 4 (PF4), interleukin(IL)-1, RANTES, beta-thromboglobulin, platelet-derived growth factor (PDGF), platelet-activating factor, CXCL7, migration inhibiting factor, thromboxane A2 and 5-HT (10). The interaction between platelets and neutrophils in the liver has been observed in models of sterile injury (24). Using intravital confocal microscopy, platelets were observed to line the endothelium at sites of focal injury facilitating neutrophil rolling through GPIIbIIIa dependent mechanisms. ACLF, often triggered by sepsis, rapidly leads to multiorgan dysfunction (6), and platelets may contribute to development of the systemic inflammatory response. In a mouse model of endotoxemia-mediated acute lung injury, P-selectin-PSGL-1 interactions were investigated (106). Administration of PSGL-1 blocking antibody reduced recruitment of neutrophils, platelet-neutrophil aggregates, lung injury and survival. It may be of interest to assess the role of this interaction in models of ALF and ACLF. Whilst platelet-neutrophil interactions may exacerbate injury, recent work may highlight a beneficial role for platelet-derived 5-HT in neutrophil recruitment to sites of inflammation. Giovanni et al. (107), demonstrate 5-HIAA, a metabolite of platelet-derived 5-HT, mediates neutrophil recruitment and transmigration to inflamed tissues via GPR35. Here, treatment with serotonin inhibitors diminished recruitment of neutrophils and clearance of peritoneal bacteria. How platelets influence neutrophil recruitment and final phenotype within the damaged liver and whether parallels from data in other organs can be extrapolated to the liver is key to develop rational antiplatelet therapies in liver disease.
Neutrophils release NETs as an antimicrobial effector mechanism (108, 109). These consist of a fibrous mesh of decondensed DNA mixed with a number of nuclear and granular proteins capturing and neutralizing microbes in an attempt to prevent dissemination (110). In the liver, NETs are released in response to infection, ischemia and sterile inflammation (111). Whilst providing an important role in pathogen defense, NETs are cytotoxic towards host cells (112). The neutrophil-platelet-NET axis is a complex interaction whereby platelet-mediated recruitment and activation of neutrophils at sites of inflammation triggers NET release (113). NET release in response to platelet activation is integrin- (114) and potentially selectin-mediated (115). Histone and polyphosphate entities within the NET matrix subsequently activate platelets through TLRs and directly activate coagulation within the bloodstream (116, 117). The neutrophil-platelet-NET axis illustrates the important role for link between infection, inflammation and thrombosis enhancing host immunity (118). In models of sepsis (116, 119), and recently in a galactosamine hydrochloride and lipopolysaccharide model of ALF (120), NETs contribute to tissue damage. Importantly, inhibition of NET generation or DNAse to breakdown NET reduces observed collateral tissue damage (116, 119). Cell-free DNA, often referred to as a NET marker (albeit somewhat non-specific) is associated with mortality in ACLF though the link with the more specific myeloperoxidase-DNA was not established (121). In a study by von Meijenfeldt et al (122), 676 patients with ALF were recruited from the U.S ALF Study group. Forty-six percent had APAP-induced ALF. Cell-free DNA and myeloperoxidase-DNA complexes were measured in comparison to healthy controls and tissue obtained at liver transplantation was stained for NETs in 20 patients. Levels of cell-free DNA and myeloperoxidase-DNA complexes were 7.1-fold and 2.5-fold higher than healthy controls respectively. High cell-free DNA was not associated with mortality. Myeloperoxidase-DNA levels were 30% higher in patients with ALF who died or required urgent liver transplant. The observed differences between ALF and ACLF may be explained, in part, by innate immune cell dysfunction in ACLF. NETs may represent an attractive therapeutic target to reduce immunothrombosis and cytotoxic cell damage, however this needs to be carefully balanced with loss of beneficial immune function. In animal models of sepsis, disruption of NET formation reduced liver injury and microcirculation thrombosis without impairing bacterial clearance (123–125). More research is required to accurately measure NET formation (108) and their impact in different models of liver injury.
Monocytes and macrophages have a critical role in homeostatic immune mechanisms, immune-mediated liver injury, fibrosis and regeneration (126). Infiltrating the site of injury 24-48hrs after neutrophils, monocytes perform diverse functions during ALF including inflammatory mediator release, clearance of dead cells, stimulation of the extracellular matrix and parenchymal regeneration (18). Distinct subsets of monocytes possess predominately inflammatory or anti-inflammatory phenotypes ( Table 1 ) (18). Differentiation into macrophages is also an important function (127, 128). Platelet interactions are evident in immune surveillance functions of the liver, initiation of inflammation, recruitment of monocytes and polarisation to a pro-inflammatory macrophage profile ( Figure 3C ) (14, 15). Monocytes, sharing common regulatory and effector properties with neutrophils, are recruited to inflamed endothelium by platelets in a similar fashion (15). Efficient monocyte recruitment requires specific stimuli, namely monocyte chemoattractant protein 1 (MCP-1) (10). The interaction between Kupffer cells and platelets is an important step in initial pathogen detection. Platelets survey macrophages through transient GPIb-vWF interactions during homeostasis (129). In the presence of blood-borne bacteria, sustained platelet-macrophage interactions are observed through vWF-GPIIb/IIIa encasing the bacterium and facilitating clearance. Increasing evidence is suggesting that the podoplanin-CLEC2 axis is central to platelet interactions with macrophages. Recently, Shan et al. (130), identified a novel interaction between platelets and macrophages potentiating APAP-induced liver injury through chitinase-3 like protein-1 (CHI3L1). CHI3L1 is a soluble protein released by multiple immune cells and found to be raised in a range of liver disease (131). In this study, CHI3L1 interaction via CD44 on macrophages upregulated podoplanin expression and subsequent platelet aggregation via CLEC2. In this model, disruption of the pathway at the level of CHI3L1 and podoplanin-CLEC2 greatly inhibited liver injury after APAP administration. In the model of APAP-induced liver injury (132), platelet depletion greatly reduced tissue damage and the study by Shan et al (130), highlights one potential pathway underlying this – its role in the other models of liver injury are yet to be determined.
Monocytes participate in early stages of the innate immune response to acute liver injury through cytokine production, antigen presentation and polarisation to inflammatory macrophages (30). Plasticity in response to the local microenvironment is demonstrated with anti-inflammatory monocytes appearing 12-24 hours later promoting resolution via IL-10, transforming growth factor beta (TGF-β) and vascular endothelial growth factor (VEGF) (133–136). This arises from recruitment and in situ reprogramming (136). Apart from monocyte differentiation, specific macrophage populations resident to the peritoneal cavity, characterised by GATA6 expression, are recruited to the inflamed liver to assist in liver repair (137–140). This appears to be mediated by ATP release and exposed hyaluronan (137, 138). Recently, Jin et al. (139), demonstrated through dual recombinase mediated genetic GATA6+ lineage tracing, macrophages are only recruited to surface of liver during carbon tetrachloride-induced liver injury, questioning a potential role in ALF. In ACLF, there is evidence of dampened function, characterised by reduced human leukocyte antigen (HLA)-DR expression, correlated with high mortality rates and increased prothrombin time (31). Underlying immunoparesis in ACLF, there is also expansion of several monocyte populations including MERTK+ (32), monocytic myeloid-derived suppressor cells (141) and intermediate CD14++CD16+ (142) with classical monocytes also exhibiting impaired function, characterised by reduced TLR2/4 expression, phagocytic activity and upregulation of genes related to dampened immune response (142). Monocyte/macrophage polarisation and plasticity is influenced by platelet activity. In vitro studies by Lee et al. (143), demonstrated adenosine diphosphate-activated platelets induced CD16 expression on CD14+CD16- monocytes from platelet-derived TGF-β and monocyte-derived IL-6. These monocytes preferentially differentiated towards M2 macrophages expressing CD163 and MerTK. It may be postulated that in platelet-monocyte interactions in advanced cirrhosis contributes to immunoparesis through expansion of MerTK macrophages precipitating widespread inflammation in ACLF. In contrast, in vitro lipopolysaccharide-treated monocytes co-incubated with platelets are skewed from an M2 towards a pro-inflammatory M1 phenotype demonstrating increased TNF-α expression, improved bacterial phagocytic activity and reduced healing capability (48). In vivo platelet transfusion increased inducible nitrous oxide synthase-expressing macrophages, improving bacterial clearance and survival in septic mice. In both of these settings, blockade of CD11b-GPIb interaction abolished the effect. The effect of platelet-macrophage interactions in liver disease is not limited to the hepatic environment as in a murine model of acute liver injury monocyte-platelet aggregates modulated microglial activation and drove the development of sickness behaviors in TLR4-dependent pathways (140). Platelet interactions with monocytes and macrophages are complex and diverse. Given the distinct cellular niches that exist within fibrotic livers (144), studying spatio-temporal platelet driven immunothrombosis in acute and chronic liver disease and how this influences macrophage phenotype and function in liver inflammation remains an exciting avenue to study.
Platelets play a provocative role in the resolution of inflammation with the ability to potentiate immune-mediated damage, promote regeneration and drive fibrosis ( Figure 4 ). This complex relationship may reflect limitations in models used to study liver inflammation, represent gaps in our knowledge or identify roles for platelets that vary throughout and in different types of injury. During liver regeneration, hepatocyte proliferation is controlled by a multitude of extracellular signals including cytokine, growth factor and metabolic pathways (15). TNF-α and IL-6 are cytokines central to regulation of liver regeneration whilst growth factors such as hepatocyte growth factor, endothelial growth factor, Insulin-like growth factor-1 and PDGF drive cell cycle progression (145). Platelets accumulate within the space of Disse rapidly after partial hepatectomy and have an active role in hepatic regeneration (146–148). Such a role for platelets is also appreciated in the clinical setting in a recent meta-analysis of 3966 patients (149). In this study preoperative thrombocytopenia constituted a significant risk factor for post-hepatectomy liver failure. Platelets mediate regeneration through interactions with LSECs, Kupffer cells and hepatocytes. These interactions are facilitated by direct contact and platelet-derived soluble mediators including hepatocyte growth factor, insulin-like growth factor 1, PDGF, VEGF, 5-HT, adenosine diphosphate and ATP (150). Downstream cascades key to these pathways include TNF-α/NF-kB, Il-6/STAT3, phosphatidylinositol 3-kinase (PI3K)/Akt and ERK1/2 (15, 151). LSECs produce mitotic substances, specifically IL-6, hepatocyte growth factor and VEGF (152). Direct adhesion between platelets and LSECs induces IL-6 release, in turn, leading to hepatocyte growth factor secretion from hepatic stellate cells promoting hepatocyte regeneration in vitro (153). This is likely controlled through podoplanin-CLEC2 signalling (154) and expression of vWF (155). Soluble mediators released by activated platelets including TGF–β1 (156) and sphingosine-1-phosphate (157) are also sufficient to support IL-6 production by LSECs. Kupffer cells are an important source of regenerative cytokines TNF-α, IL-6 and IL-1β (126). Depletion of Kupffer cells impairs hepatocyte proliferation during liver regeneration in cholestatic injury (158), alcohol-induced injury (159) and partial hepatectomy (147). Platelets contribute to regeneration through promoting TNF-α production by Kupffer cells, however utilize alternate pathways in Kupffer cell depleted environments (147). Platelet-derived 5-HT is an important mediator of liver regeneration, most likely through production of growth factors at the site of injury (160, 161). Recently a multi-center trial has demonstrated that perioperative use of selective serotonin reuptake inhibitors and serotonin noradrenaline reuptake inhibitors is associated with adverse outcomes after hepatic resection further supporting a role for 5-HT (162). Within the space of Disse, platelets are also able to interact with hepatocytes directly. Internalization of platelets and platelet-like particles followed by horizontal transfer of mRNA has also been demonstrated to contribute to hepatocyte proliferation (163). Linking coagulation and inflammation, liver-specific tissue factor release promotes platelet accumulation through fibrin(ogen) deposition, facilitating resolution after partial hepatectomy (164). Whilst it is generally accepted that platelets are able to promote liver regeneration through a variety of mechanisms, there is emerging evidence implicating them in delaying resolution (12, 77). Neutrophils are not only central drivers of inflammation but also promote resolution during sterile injury (24). Recently our group demonstrated signalling via podoplanin-CLEC2 between platelets and inflammatory macrophages reduced TNF-α secretion and subsequent neutrophil recruitment to facilitate resolution of inflammation in APAP toxicity (12). Moreover, increased vWF deposition and impaired clearance has been attributed to persisting platelet accumulation in APAP-induced liver disease (77). In this model, deficiency or inhibition of vWF accelerated resolution. These results add a further layer of complexity to our understanding highlighting that targeting platelets may be temporally sensitive but also vary in different models of liver injury (12, 77, 155). Partial hepatectomy is the most commonly studied model of liver regeneration, however pathways involved in injury and resolution vary in other models (165). Thus, further investigation is required in different models of ALF and ACLF to elucidate specific roles of platelets and identify potential therapeutic targets.
Platelets play a vital role at all stages of acute liver injury through direct and indirect interactions with immune cells, stromal cells and the endothelium. Their involvement is both beneficial and detrimental. On one hand, they can intelligently sense and appropriately respond to pathogens, recruit leucocytes and promote regeneration at the resolution of inflammation. On the other, they exaggerate immune-mediated tissue injury, worsen hepatocyte dysfunction through microthrombi formation and delay mechanisms of resolution. Pathways controlling platelet effector function, such as the podoplanin-CLEC2 axis, represent an attractive therapeutic target however the disease-specific and temporal roles of platelets need to be carefully dissected in order to develop effective disease modifying treatments.
SM – conceptualization, writing of original draft and editing. AC – overall supervision of article. All authors contributed to the article and approved the submitted version.
All figures created with BioRender.com
Author AC is recipient of a Wellcome trust clinical fellowship and an NIHR clinical lectureship. AC has received consultation fees from Principia biopharma now part of Sanofi. The remaining author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647049 | Yufei Xu,Fengli Zuo,Huiling Wang,Jing Jing,Xiujing He | The current landscape of predictive and prognostic biomarkers for immune checkpoint blockade in ovarian cancer | 27-10-2022 | ovarian cancer,immune checkpoint blockade,biomarker,immunotherapy response,prognosis | Immune checkpoint blockade (ICB) therapy has evoked a prominent shift in anticancer therapy. Durable clinical antitumor activity to ICB has been observed in patients with ovarian cancer (OC). However, only a subset of patients derive clinical benefit, and immune-related adverse events (irAEs) caused by ICB therapy can lead to permanent tissue damage and even fatal consequences. It is thus urgent to develop predictive biomarkers to optimize patient outcomes and minimize toxicity risk. Herein, we review current predictive and prognostic biomarkers for checkpoint immunotherapy in OC and highlight emerging biomarkers to guide treatment with ICB. The prevalent biomarkers, such as PD-L1 expression status, tumor-infiltrating lymphocytes, mutational burden, and immune gene signatures, are further discussed. We provide a state-of-the-art survey on prognostic and predictive biomarkers for checkpoint immunotherapy and offer valuable information for guiding precision immunotherapy | The current landscape of predictive and prognostic biomarkers for immune checkpoint blockade in ovarian cancer
Immune checkpoint blockade (ICB) therapy has evoked a prominent shift in anticancer therapy. Durable clinical antitumor activity to ICB has been observed in patients with ovarian cancer (OC). However, only a subset of patients derive clinical benefit, and immune-related adverse events (irAEs) caused by ICB therapy can lead to permanent tissue damage and even fatal consequences. It is thus urgent to develop predictive biomarkers to optimize patient outcomes and minimize toxicity risk. Herein, we review current predictive and prognostic biomarkers for checkpoint immunotherapy in OC and highlight emerging biomarkers to guide treatment with ICB. The prevalent biomarkers, such as PD-L1 expression status, tumor-infiltrating lymphocytes, mutational burden, and immune gene signatures, are further discussed. We provide a state-of-the-art survey on prognostic and predictive biomarkers for checkpoint immunotherapy and offer valuable information for guiding precision immunotherapy
Immune checkpoint blockade therapies (ICBs) can circumvent tumor-mediated immune suppression and reinvigorate antitumor immune responses, in contrast with conventional therapeutic strategies that exert direct cytotoxicity against tumor cells (1, 2). Immune checkpoint inhibitors (ICIs) that target the programmed cell death protein-1 (PD-1)/programmed death receptor ligand-1 (PD-L1) axis or cytotoxic T lymphocyte antigen 4 (CTLA4) have achieved impressive success against various cancer types (3). ICIs have achieved remarkable clinical activity with durable disease control across multiple advanced tumors (4). Accordingly, several ICIs have been approved by the United States Food and Drug Administration (FDA) for patients with malignancies, including melanoma, lung cancer, triple-negative breast cancer (TNBC), colorectal cancer, gastric cancer, renal cell cancer, head and neck squamous cell cancer, bladder cancer, lymphoma and so on (5). Albeit substantial advancements in clinical therapy, only a minority of patients receiving ICIs derive benefits. In addition, ICB therapy is significantly restricted by the occurrence of immune-related adverse events (irAEs), resulting from immune hyperactivation and subsequent immune homeostasis disturbance. Severe adverse events can lead to permanent disorders and can be lethal in some cases (6). Therefore, there is intense interest in developing predictive and prognostic biomarkers for ICI therapy to better understand the benefits and risks driven by ICB and effectively select patients. Manipulating the immune environment with ICIs is an attractive therapeutic approach for antitumor therapy in ovarian cancer (OC) ( Figure 1 ). There has been considerable progress in utilizing ICB therapy for OC over the past few years ( Table 1 ; Supplementary Table S1 ). However, there is still confusion regarding patient selection and the choice of therapeutic regimen for patients with OC, underscoring the need for effective biomarkers to predict response and remission. In this review, we attempt to summarize published original research and clinical trials involving biomarker assessment in OC receiving ICI therapy and discuss ongoing efforts to develop predictive biomarkers of responsiveness and outcomes.
Direct measurement of PD-L1 expression is a logical biomarker for predicting response to anti-PD-1/PD-L1 therapies. PD-L1 immunohistochemistry (IHC) assay is now FDA-approved as a companion diagnostic biomarker to select patients most likely to benefit from ICI treatment for multiple cancer types, such as non-small cell lung cancer(NSCLC), metastatic TNBC, and melanoma (5). The predictive value of PD-L1 expression was assessed in OC patients treated with anti-PD-1/PD-L1 antibodies ( Table 1 ). KEYNOTE-100 (NCT02674061) investigated the clinical activity of pembrolizumab in patients with recurrent advanced OC and introduced PD-L1 stain score as a predictive biomarker, in which patients with higher PD-L1 expression (combined positive score≥10) had an increased overall response rate (ORR) and prolonged overall survival (OS) with pembrolizumab (8). More recently, Sanborn et al. evaluated the efficacy and safety of varlilumab plus nivolumab in patients with advanced solid tumors (10). Significantly, an absolute increase of 5% or more in tumor PD-L1 expression induced by treatment tended to improve progression-free survival (PFS) in OC (7.4 months vs. 3.5 months, p=0.07), whereas baseline pretreatment PD-L1 expression was not associated with ORR (10). Prespecified biomarker analysis in the JAVELIN-200 trial revealed a trend for prolonged PFS with the addition of avelumab to pegylated liposomal doxorubicin (PLD) compared with PLD alone among OC patients with PD-L1-positive tumors (12). Nevertheless, several trials yielded inconsistent or even contradictory results regarding the role of PD-L1 expression as a marker for predicting response to ICB and clinical outcomes in OC. Liu et al. (15) obtained the opposite results in evaluating the predictive and prognostic value of PD-L1 expression in recurrent OC patients receiving nivolumab and bevacizumab. Even patients with PD-L1-negative tumors (10/22) had higher therapeutic activity than those with PD-L1-positive expression (2/14) (15). In addition, several studies have shown that the expression of PD-L1 was not predictive of ICI outcome and prognosis in OC patients (36, 38–43). Potential reasons for these paradoxical results include the inability to accurately reflect PD-L1 status due to PD-L1 expression transiency and heterogeneity, differences in the disease status of patients, the poor uniformity between various detection assays, and the lack of standardized criteria and thresholds for assessing positivity (3, 44, 45). Therefore, PD-L1 status is likely insufficient to determine the suitability of ICI therapy for OC patients. Further refinement of the use of PD-L1 expression status as a robust biomarker for checkpoint immunotherapy is warranted.
TIICs can serve as an index to monitor the tumor microenvironment (TME) and play an increasingly important role in the immune response against cancer (46). Therefore, TIICs have also been speculated to be surrogate biomarkers for ICB immunotherapy in many types of cancer, including OC ( Table 1 ). A comprehensive analysis of immune cells in patients with epithelial ovarian cancer (EOC) revealed a positive correlation between the infiltration of immune cells and the clinical outcome of EOC (16). The density of tumor-infiltrating lymphocytes (TILs), specifically CD8+ T cells, is a solid positive prognostic indicator for multiple cancer types regardless of ICI therapy. In fact, CD8 expression in tumors was predictive of clinical benefit with avelumab plus PLD treatment in OC (12). Of note, patients with dual PD-L1-positive and CD8-positive tumors seemed to benefit more from combination treatment than subgroups defined by only one of these biomarkers (12). Another potential predictor of ICI response is tumor-infiltrating mast cells (TIMs) within a tumor ( Table 1 ). In high-grade plasmacytoid ovarian cancer (HGSOC), stromal TIMs (sTIMs) abundance was negatively associated with the ICB response (18). Remarkably, tumors with low sTIMs had enhanced effector functions of CD8+ T cells (18). This finding was corroborated in short-term HGSOC organoids. The effector molecules (GZMB and IFN-γ) on CD8+ T cells were marginally increased in organoids derived from low sTIMs tumors, compared to organoids from high sTIMs tumors (18). Overall, the abundance of sTIMs predicts a dismal prognosis in HGSOC patients treated with anti-PD-1 therapy. Except for the spatial position and density of TIICs, their phenotype and activation status also impact the clinical benefit of ICIs (3). The immune-inflamed phenotype is usually accompanied by the expression of PD-L1 on infiltrating immune cells and tumor cells, which is associated with a better response to ICI therapy (3). In a trial investigating combination regimens with anti-PD-L1 antibody in women’s cancers, a trend toward a positive association of treatment response with the degree of PD-L1-positive TILs was observed (39). In contrast, melanoma patients with PD-L1-positive TILs had a significantly worse prognosis than those with PD-L1-negative TILs (P = 0.008) (47). Further investigations are needed to determine whether PD-L1-positive TILs are suitable to serve as predictors of ICB effectiveness. In addition, other non-neoplastic cells in the TME are also non-negligible, which are probably of biological significance. Therefore, increased awareness of the role of these distinct TME compartments is needed for comprehensive biomarker development to predict ICB response and prognosis.
Tumor development and progression generally occur along with the acquisition and accumulation of mutations (45). Neoantigens generated by mutations may lead to T-cell infiltration, thereby better response to immunotherapy (48). In fact, several studies have attempted to evaluate somatic mutations as biomarkers for predicting ICB response in OC ( Table 1 ). ARID1A mutation or loss was associated with immune microenvironmental factors in clear cell ovarian cancer (CCC), suggesting that ARID1A status has potential as a biomarker to guide decisions concerning patient selection for ICB therapy in CCC (20). The phase I/II trial (NCT02657889) reported two novel biomarkers for the combination of poly (adenosine diphosphate-ribose) polymerase (PARP) and PD-1 inhibitors in the treatment of platinum-resistant OC (17). Mutational signature 3 reflected homologous recombination deficiency (HRD) status, and positive immune score (IS) was a surrogate of interferon-primed exhausted CD8+ T cells in TME. Specifically, the presence of one or both of the above alternative markers was associated with significantly prolonged PFS (HR = 0.32), while concurrent absence showed no response to PARP/PD-1 inhibitors(ORR= 0%) (17). Another metric, known as tumor mutation burden (TMB), is a strong predictor of ICB efficacy. Unfortunately, its predictive performance in OC is disappointing. No significant correlation was found between TMB and immunotherapy response in recurrent OC (21). Furthermore, BRCA1/2 mutations and HRD status also did not predict the clinical benefit of ICI in heavily pretreated patients with OC (21). Notably, additional exploratory analyses identified the fraction of genome altered (FGA) as a promising biomarker of response to ICI in OC, which can characterize global copy number alterations. High FGA was significantly associated with improved OS (HR = 0.49; log-rank P = 0.01) and PFS (HR = 0.54; log-rank P = 0.014) after ICI therapy in OC (21). The optimal cutoff for defining high vs. low FGA is unclear; therefore, the predictive capacity of FGA warrants further validation. TMB was also explored in the phase I/II trial (NCT03029598), which evaluated pembrolizumab and carboplatin for recurrent or refractory ovarian, fallopian tube, or primary peritoneal cancer (19). Stratification by the ratio of peripheral CD8+PD1+Ki67+ T cells to tumor burden at baseline yielded a significant survival advantage. Patients with a low ratio (<0.0375) had a median OS of only 8.72 months, while those with a high ratio (≥0.0375) had a significantly longer median OS of 18.37 months (p=0.0099). However, no significant survival difference was observed when using CD8+PD1+Ki67+ T cell (p=0.53) or tumor burden alone (p=0.24) as stratification criteria (19). Overall, TMB alone does not clearly discriminate responders from non-responders in OC patients treated with ICIs.
Gene expression analysis can uncover global tumor and microenvironment features, providing promise for predicting the clinical benefit of checkpoint inhibitor strategies. Multiplex characterization of the TME and gene expression signatures have been proposed as effective methods to dissect the immune contexture and cancer cell-intrinsic features. According to TME information derived from transcriptome data of OC, Li et al. (23) established immune cell infiltration (ICI) scores and an immune-related gene prognostic model to predict the clinical benefits of OC patients undergoing immunotherapy. Signal transducer and activator of transcription 1 (STAT1) has been demonstrated to be associated with TME. A recent study found that STAT1 expression was positively correlated with PD-L1 expression and had the potential to predict the response to ICB in patients with EOC (24). Integrins are transmembrane receptors that mediate the connection between cells and their external environment (49–51). Several immune-related gene signatures have been confirmed to predict the immunotherapeutic response in OC. The TGF-β regulated signaling pathway was noted to contribute to immunotherapy resistance in OC (27). A significant negative correlation between the TGF-β score and ICI-PFS was observed in OC, with an ICI-PFS of 16.6 months in the low TGF-β score group compared to 2.65 months in the high TGF-β score group (p = 0.0012). As the most common RNA modification, N6-methyladenosine (m6A) plays a key role in epigenetics (52). A risk model based on m6A-related targets has an excellent clinical prognostic stratification effect in advanced OC. Importantly, the high- and low-risk groups divided by this model have significant differences in TME contexture, suggesting that this model may be able to predict immunotherapy response in OC (29). Chemokines have essential roles in modulating immune homeostasis and inflammatory responses (53). Accumulating findings suggest that chemokines can influence cancer cell proliferation, invasion, angiogenesis, and therapy resistance by recruiting immune cells and modulating the TME (54, 55). The prognostic and predictive values of the CXC chemokine family have been addressed in the setting of OC, including CXCL9, CXCL11, and CXCL13 ( Table 1 ). Tumors with high CXCL9 expression had significantly prolonged OS, implying the feasibility of CXCL9 expression as a novel prognostic marker for high-grade serous ovarian cancer (HGSC) (30). Similarly, Fan et al. (33) found a significant positive correlation between the expression of CXCL13, FCRLA, PLA2G2D, and MS4A1 and a better prognosis of OC. Meanwhile, these potential therapeutic genes could reflect OC immune status and allow better predictions of who will respond to ICI. Furthermore, Yang etal. (32) examined the therapeutic effects of CXCL13 and PD-1 blockade in human HGSC tumors and mouse models. They found that CXCL13 can augment the efficacy of PD-1 checkpoint blockade in HGSC by shaping the antitumor microenvironment. CXCL13 can facilitate CXCR5+CD8+ T-cell recruitment to tertiary lymphoid structures. Furthermore, the combination of CXCL13, CD8, and CXCR5 was confirmed as a potential prognostic indicator or response biomarker for ICB therapy in patients with HGSC. CXCL11 expression has been demonstrated as a biomarker for predicting the response to anti-PD-1/PD-L1 therapy in a clinical trial of OC (31). In OC patients with HRD, tumors with high CXCL11 expression had a more robust immune response to PD-L1 blockade than those with low CXCL11 expression. Notably, the tumor-infiltrating immunophenotype and neoantigen burden were significantly elevated in CXCL11-high tumors. In addition, several genes have been demonstrated to be associated with immunotherapy efficacy and prognosis in OC ( Table 1 ). For example, Capping Actin Protein, Gelsolin-Like (CAPG) (25) and Layilin (LAYN) (26) appeared to be indicators of ICI outcome. Tumors with high CAPG or LAYN expression showed a significantly shorter survival time. In a study, the predictive significance of NAD+ metabolism-related genes (NMRGs) on immunotherapy response in patients with OC was examined. The high-risk score obtained by the NMRG-based model was also associated with a poorer prognosis (28). Apolipoprotein B mRNA editing enzyme catalytic subunit 3A (APOBEC3A) has been recognized as an indicator of genomic instability and may aid in predicting the prognosis and response to immunotherapy in OC (22).
In recent years, there has been great interest in developing blood-derived predictive biomarkers of ICI response, owing to its convenient and non-invasive sampling (56). Cancer antigen 125 (CA-125) is an important tumor biomarker specific to OC (57); thus, several studies have carried out exploratory research on the predictive role of CA-125 in OC patients treated with ICIs ( Table 1 ). A phase II trial (NCT02608684), designed for evaluating the combination of pembrolizumab and chemotherapy in platinum-resistant OC, found CA-125 to be a reliable marker that reflected response and progression (42). In a retrospective study of EOC patients treated with ICI (35), the magnitude of increase in CA-125 levels within the first 12 weeks of treatment was significantly smaller in patients with clinical benefit than in those without benefit, suggesting a possible predictive role for the degree of CA-125 increase. In a phase 1b study of avelumab in patients with heavily pretreated OC, 12 patients with an objective response, of whom all 7 patients evaluable for CA-125 levels showed decreased CA-125 concentrations (36). Dynamic monitoring of circulating tumor DNA (ctDNA) in plasma samples offers a meaningful direction for biomarker identification for immunotherapy in OC patients (37). A satisfying finding was that ctDNA concentration was related to clinical response and benefit, although the effect sizes were modest (37). Additionally, in a phase II trial of olaparib combined with durvalumab for OC, increased IFNγ production and elevated VEGFR3 levels in blood samples showed positive and negative correlations with PFS, respectively (p=0.023; p=0.017) (34).
The clinical trials and original research outlined above have shown that classical biomarkers derived from the TME and tumor intrinsic features, such as PD-L1 expression, TMB, TIICs, and transcriptomic signatures, were correlated with ICI response and outcome in OC. Although these findings are intriguing, the implementation of these classical biomarkers has been hampered by inconsistencies and limitations. Promisingly, new biomarkers often designed as substitutes or complements to conventional biomarkers are constantly emerging, such as microbiome, tertiary lymphoid structures (TLSs), and tumor-associated antigens (TAAs). The potential of microbiome and its derived metabolome as biomarkers for predicting the efficacy of immunotherapy has been validated in melanoma (58), lung cancer (59), hepatobiliary cancer (60), and colorectal cancer (61). Several studies have demonstrated that clinical outcomes of immunotherapy for solid tumors are strongly correlated with the presence of TLSs, suggesting that TLSs may be a valid predictive indicator in the future (62). Elevated levels of carcinoembryonic antigen (CEA) have also been reported to negatively correlate with the prognosis of resected NSCLC patients receiving ICB therapy (63). More recently, a comprehensive predictive model for ICB response was developed across 16 different cancer types, which included the features of peripheral blood such as platelets, neutrophil-to-lymphocyte ratio, albumin, and hemoglobin (HGB) (64). These studies provide new perspectives to develop new biomarkers for OC patients treated with ICB therapy. The predictive values of these biomarkers in OC remain to be validated in routine clinical settings. As evidenced by the fact that a single biomarker is often insufficient to determine the suitability of ICI therapy for OC patients, the combination of different biomarkers may be more valuable in predicting the clinical prognosis and therapeutic response to immunotherapy. Indeed, it has been proposed that the incorporation of dynamic and static biomarkers could improve decision-making to design tailored immunotherapy strategies. Moreover, the development of relevant biomarkers for the toxicity prediction of ICB therapy has become a research hotspot and is expected to offer effective ways to uncouple immunotherapy toxicity from its antitumor activity.
YX and XH: conceptualization and writing-original draft preparation. FZ: visualization. YX, FZ, HW, JJ and XH: writing-review and editing. XH and JJ: supervision and funding acquisition. All authors have read and agree to the published version of the manuscript.
This work was supported by (1) National Natural Science Foundation of China (No. 82172634 and 81902792); (2) Key Program of the Science and Technology Bureau of Sichuan (No. 2021YFSY0007); (3) 1.3.5 project for disciplines of excellence, West China Hospital, Sichuan University (No. ZYYC20013).
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647051 | Xiang Zhong,Shu Xu,Quhui Wang,Long Peng,Feiran Wang,Tianyi He,Changyue Liu,Sujie Ni,Zhixian He | CAPN8 involves with exhausted, inflamed, and desert immune microenvironment to influence the metastasis of thyroid cancer | 27-10-2022 | CAPN8,thyroid cancer,prognosis,immunotherapy,tumor immune microenvironment | Background Thyroid cancer (THCA) is the most prevalent malignant disease of the endocrine system, in which 5-year survival can attain about 95%, but patients with metastasis have a poor prognosis. Very little is known about the role of CAPN8 in the metastasis of THCA. In particular, the effect of CAPN8 on the tumor immune microenvironment (TIME) and immunotherapy response is unclear. Material and methods Multiome datasets and multiple cohorts were acquired for analysis. Firstly, the expression and the prognostic value of CAPN8 were explored in public datasets and in vitro tumor tissues. Then, hierarchical clustering analysis was performed to identify the immune subtypes of THCA according to the expression of CAPN8 and the activities of related pathways. Subsequent analyses explored the different patterns of TIME, genetic alteration, DNA replication stress, drug sensitivity, and immunotherapy response among the three immune phenotypes. Finally, five individual cohorts of thyroid cancer were utilized to test the robustness and extrapolation of the three immune clusters. Results CAPN8 was found to be a significant risk factor for THCA with a markedly elevated level of mRNA and protein in tumor tissues. This potential oncogene could induce the activation of epithelial–mesenchymal transition and E2F-targeted pathways. Three subtypes were identified for THCA, including immune exhausted, inflamed, and immune desert phenotypes. The exhausted type was characterized by a markedly increased expression of inhibitory receptors and infiltration of immune cells but was much more likely to respond to immunotherapy. The immune desert type was resistant to common chemotherapeutics with extensive genomic mutation and copy number variance. Conclusion The present study firstly explored the role of CAPN8 in the metastasis of THCA from the aspects of TIME. Three immune subtypes were identified with quite different patterns of prognosis, immunotherapy response, and drug sensitivity, providing novel insights for the treatment of THCA and helping understand the cross-talk between CAPN8 and tumor immune microenvironment. | CAPN8 involves with exhausted, inflamed, and desert immune microenvironment to influence the metastasis of thyroid cancer
Thyroid cancer (THCA) is the most prevalent malignant disease of the endocrine system, in which 5-year survival can attain about 95%, but patients with metastasis have a poor prognosis. Very little is known about the role of CAPN8 in the metastasis of THCA. In particular, the effect of CAPN8 on the tumor immune microenvironment (TIME) and immunotherapy response is unclear.
Multiome datasets and multiple cohorts were acquired for analysis. Firstly, the expression and the prognostic value of CAPN8 were explored in public datasets and in vitro tumor tissues. Then, hierarchical clustering analysis was performed to identify the immune subtypes of THCA according to the expression of CAPN8 and the activities of related pathways. Subsequent analyses explored the different patterns of TIME, genetic alteration, DNA replication stress, drug sensitivity, and immunotherapy response among the three immune phenotypes. Finally, five individual cohorts of thyroid cancer were utilized to test the robustness and extrapolation of the three immune clusters.
CAPN8 was found to be a significant risk factor for THCA with a markedly elevated level of mRNA and protein in tumor tissues. This potential oncogene could induce the activation of epithelial–mesenchymal transition and E2F-targeted pathways. Three subtypes were identified for THCA, including immune exhausted, inflamed, and immune desert phenotypes. The exhausted type was characterized by a markedly increased expression of inhibitory receptors and infiltration of immune cells but was much more likely to respond to immunotherapy. The immune desert type was resistant to common chemotherapeutics with extensive genomic mutation and copy number variance.
The present study firstly explored the role of CAPN8 in the metastasis of THCA from the aspects of TIME. Three immune subtypes were identified with quite different patterns of prognosis, immunotherapy response, and drug sensitivity, providing novel insights for the treatment of THCA and helping understand the cross-talk between CAPN8 and tumor immune microenvironment.
Thyroid carcinoma (THCA) is the most prevalent malignant disease of the endocrine system, which can be divided into four histological types, including papillary thyroid cancer (PTC), follicular thyroid cancer (FTC), medullary thyroid cancer, and poorly differentiated thyroid cancer (1). The 5-year survival rate for patients with PTC or FTC can attain about 95%, but patients with metastatic THCA have a poor prognosis (2). Calpain calcium kinase (CAPN) is a kind of cysteine protein kinase widely existing in most eukaryotic cells and plays a key role in regulating cell cycle and apoptosis (3). It is already reported that the aberrant expression of CAPN is involved in several types of cancer progression by inducing NF-κB, focal adhesion kinase, and MYC pathways (4–7). However, very little is known about the role of CAPN8 in the genesis and development of THCA up to now. In particular, the effect of CAPN8 on the tumor immune microenvironment (TIME), which is a well-recognized factor in promoting the metastasis of THCA, is unclear (8). Hereby we hypothesize that CAPN8 might facilitate the metastasis of thyroid cancer cells and lead to poor prognosis by inducing an inhibitory TIME pattern. Firstly, we explored the expression and downstream signaling pathways of CAPN8 in The Cancer Genome Atlas—Thyroid Cancer (TCGA-THCA) cohort and in vitro tumor tissues. Next, clustering analysis was performed, and three immune-related clusters (immune exhausted, immune desert, and inflamed) were identified for THCA according to the expression of CAPN8 and related pathways. Subsequent analyses examined the different patterns of genetic alteration, DNA replication stress, TIME, immunotherapy response, drug sensitivity, and prognosis amid the three immune clusters of THCA. Finally, external validation cohorts were utilized to test the robustness and extrapolation of the three immune clusters. Overall, the present study is aimed at elucidating the role of CAPN8 in the metastasis of THCA from the aspects of TIME, DNA replication stress, and genetic variation. These findings will provide novel insights for the treatment of THCA and help understand the cross-talk between CAPN8 and the tumor immune microenvironment.
Multiome datasets of thyroid cancer were obtained from TCGA-THCA (497 tumor samples and 71 normal samples) (9). RNA-seq data, downloaded in the format of fragments per kilobase million at the UCSC Xena website (9), was transformed into the value of transcripts per kilobase million for further analysis. Information about copy number variance (CNV) was acquired from the FireBrowse (10) data portal. Detailed somatic mutation categories were retrieved from the cBioPortal (11) online platform. Meanwhile, five datasets of thyroid cancer were also exported from the Gene Expression Omnibus database for external validation, including GSE3467 (n = 9), GSE3678 (n = 7), GSE33630 (n = 49), GSE60542 (n = 33), and GSE27155 (n = 95) cohorts. The batch effect amid different arrays was eliminated by using the ComBat function of R (version 4.1.3) package sva (12). As these data are open access resources from public database where patients’ consents were already obtained, extra informed consents are not needed.
Firstly, the number of positive cells and the total cells in each stained section were counted to calculate the positive rate (PP) (PP% = positive cells/total cells). By averaging the PP values of 10 discontinuous fields of the experimental tissue in a microscope with a 200-fold high-power lens, the patients were scored with 0 point for no positive cells and 1, 2, 3, and 4 points for 0% < PP ≤ 10%, 11% ≤ PP <50%, 50% ≤ PP < 80%, and PP ≥ 80%, respectively. Then, the staining intensity (SI) of cells in the tissue was estimated based on the shades of cell color. The SI score was marked as 0 when there was no obvious staining and 1, 2, and 3 for light brownish yellow, brownish yellow, and brown staining, respectively. The final IRS score was calculated by the following formula: IRS = PP × SI. IRS >3 indicates a high expression, while IRS ≤3 represents a low expression.
The expression difference of CAPN8 between 33 types of cancer and cancer-adjacent tissues was illustrated in a boxplot by using the UCSC Xena web browser. Meanwhile, immunohistochemistry (IHC) was also performed on isolated thyroid cancer tissues to detect the level of CAPN8 protein. IHC was conducted as described previously (13), and tumor sections were obtained from patients who had received radical surgery for thyroid carcinoma in the Affiliated Hospital of Nantong University. The primary anti-bodies used for IHC were anti-CAPN8 (1:30, biorbyt, orb140072).
To test the prognostic value of CAPN8, 496 patients in the TCGA-THCA cohort were divided into CAPN8-high and CAPN8-low groups according to the median mRNA value. The Kaplan–Meier (K-m) curve and log-rank test were then utilized to show their difference in progression-free survival (PFS) time. To explore the biological function of CAPN8, differential expression analysis was carried out between CAPN8-high and CAPN8-low groups by using R package limma (14). |Log2 fold change (FC)| >1 and false discovery rate (FDR) <0.05 were set as the significant threshold. Subsequently, Gene Set Enrichment Analysis (GSEA) (15) was performed to recognize the differentially expressed pathways (DEPs) between CAPN8-high and CAPN8-low groups by using R package clusterProfiler. In total, 50 well-known cancer hallmarks (16) were set as the background gene sets, and FDR <0.05 was chosen as the significant threshold.
To elucidate the potential regulating mechanism of CAPN8 toward thyroid cancer cells, Least Absolute Selection and Shrinkage Operator (LASSO) penalty and ridge regression were implemented to screen the 22 significant DEPs by using R package glmnet (17). In addition, random survival forest (RSF) algorithm was also employed to compute the significance of each DEP by using the R packages randomForestSRC and randomSurvivalForest (https://CRAN.R-project.org/package=beeswarm) with the minimal depth method to determine the final number of prognostic variables. Pathways with a certain contribution to patients’ overall survival were screened in the two models. The importance of each variable was then visualized in a bar plot, and the marginal effect was displayed by the function plot.error of R package randomSurvivalForest.
The six DEPs, obtained by adaptive LASSO regression, and eight DEPs, obtained by random survival forest, were taken into intersection with a Venn diagram depicting the common DEPs by using R package VeenDiagram (18). Gene Set Variation Analysis (GSVA) (15) was then performed to quantify the pathway activities of three common DPEs in THCA. Afterwards, survival analysis was conducted to demonstrate the impact of the three DEPs on PFS. The patients were stratified into two groups according to the median activity score of each DEP, with the K-m curve showing their difference in PFS. Of the common DEPs, HALLMARK_E2F_TARGETS was a prominent risk factor for THCA patients. Therefore, the core enrichment genes of HALLMARK_E2F_TARGETS were then submitted to hierarchical clustering analysis to identify the subtypes of THCA, resulting in an E2F-Clust with two sub-clusters. Hierarchical clustering was completed with Ward’s Clustering, computing the Euclidean distance among each patient by using R function hclust. Consensus Cumulative Distribution Function (CDF) and Delta area (relative change of area under the CDF curve) were used to select the proper clustering numbers. The two indices were provided in R package “ConsensusClusterPlus”.
GSEA analysis was wielded to dissect the biological features of the two E2F-Clusters of THCA, and survival analysis was conducted to explore their difference in PFS. Moreover, the infiltrating proportions of 10 immune cells were calculated for each TCHA patient by using R package MCPcounter (19) to probe the different TIME pattern (tumor immune microenvironment) between the two E2F-Clusters. In addition, the expression profiles of eight well-known immune inhibitory receptors (IRs)—CD274, PDCD1, CD247, PDCD1LG2, CTLA4, TNFRSF9, TNFRSF4, and TLR9—were also explored in the two E2F-Clusters. With these immune-related information, the two E2F-Clusters were further subdivided into three clusters (ImmCluster) by using the hclust function in R package ComplexHeatmap (20). Subsequently, the immune enrichment score (IES) and the stromal enrichment score (SES) were compared among the three ImmClusters. IES and SES were obtained by applying ESTIMATE algorithm (21) to each TCHA patient, where IES represented the enrichment score of immune cell ingredients, while SES reflected the ratio of stromal components in tumor tissues. Furthermore, the patients’ diverse responses to immunotherapy were predicted among the three ImmClusters by using Tumor Immune Dysfunction and Exclusion) algorithm (22) and R package Submap (23, 24).
To characterize the different genetic alteration profile among the two E2F-Cluster and three ImmClusters, somatic mutation, CNV, and chromosome instability for each patient were explored by R package MOVICS (Multi-Omics Integration and Visualization in Cancer Subtyping.) (25). CN GISTIC score was also computed for patients in each Immun-Cluster to manifest their different chromosome instability. Specifically, FGA, FGL, and FGG represented the fraction of CN altered genome, fraction of CN-lost genome, and CN-gained genome, respectively.
As E2F is part of the cell cycle-related pathways, which can lead to DNA replication stress and drug resistance in cancer cells, we further investigated the activities of 21 pathways related to DNA replication stress (26). Pathway activity was estimated by the GSVA strategy as previously described, and a heat map was utilized to demonstrate the difference between two E2F-Clusters and three ImmClusters. Moreover, the IC50 (half-maximal inhibitory concentration) of five typical chemotherapeutics for THCA was computed and compared between each cluster by using R package pRRophetic (27).
To extrapolate the E2F-Clust and ImmClust for further application in clinical practice, five external datasets of thyroid cancer cohorts were used for validation. One dataset was GSE27155 (n = 95) which was quantified by Affymetrix Human Genome U133A Array and annotated by GPL96 platform. The other four datasets—GSE3467 (n = 9), GSE3678 (n = 7), GSE33630 (n = 49), and GSE60542 (n = 33) were quantified by Affymetrix Human Genome U133 Plus 2.0 Array and annotated by GPL570 platform. The last four datasets were consolidated by using the ComBat function of R package sva. Principal component analysis was then utilized to visualize the homogeneity of different samples after combination. Then, based on the core enrichment genes of the E2F pathway, hierarchical clustering analysis was implemented to seek similar subclusters in these datasets. Similarly, patterns of IRs, TIME, DNA replication stress, drug resistance, and immunotherapy responses were explored in different subclusters by using the same analysis strategies as described above.
Data processing and bioinformatics analyses were accomplished by R (version 4.1.3). Packages like limma, ggplot2, survminer, clusterProfiler, GSVA, glmnet, MCPcounter, SubMap, MOVICS, etc., were employed for analyses with proper citation. Wilcox or Kruskal–Wallis tests were applied for comparisons between two or more groups involved in this study. Pearson and Spearman rank correlation were adopted to estimate the statistical correlation of parametric or non-parametric variables. A log-rank test was utilized for survival analysis. Two-sided P <0.05 was considered the significant threshold for all statistical tests.
CAPN8 showed a substantial rise in protein, mRNA, and relative IHC score (D) in thyroid cancer tissues (Figures 1A, B, D). Upon dividing the patients into two groups according to the median value of CAPN8 mRNA, patients in the CAPN8-high group were observed to have a noticeable worse survival outcome, suggesting a potential role of oncogene for CAPN8 (Figure 1C). The differential expression analysis showed that many genes showed a considerable un-regulation in the CAPN8-high group, while the levels of a relatively few genes decreased (Figure 1E). In terms of the 50 cancer hallmarks, interferon-gamma response, inflammatory response, E2F-targets, etc., were significantly upregulated in the CAPN8-high group, while the oxidative phosphorylation pathway was slightly downregulated (Figure 1F). We also detected that the strong relationship of CAPN8 with immolators were presented in the BEST website (Figure 2A), and CAPN8 could be an immunotherapy predictor for patients who underwent immunotherapy (Figure 2B).
Six and eight CAPN8-related cancer hallmarks were screened out by the adaptive LASSO regression and random survival forest algorithm, respectively (Figures 3A–C). Each pathway was graded in order of their importance, and the E2F-targets pathway showed a dominant impact on survival time in the RSF analysis (Figure 3D). In addition, the marginal effect of RSF was demonstrated in the scatter diagram where the E2F-target and G2M-checkpoint pathways exhibited a mild positive correlation with the mortality of THCA patients (Figure 3E). These findings suggest that the E2F-targets pathway could be the downstream signaling pathway of CAPN8 and plays a key role in THCA progression. Subsequently, three shared cancer hallmarks were documented after taking the intersection of six pathways from LASSO regression and eight pathways from RSF analysis, including E2F-targets, oxidative phosphorylation, and inflammatory-response pathways (Figure 4A). Of the three pathways, E2F-targets was a risk factor, and patients with a high pathway activity of E2F-targets demonstrated a worse survival outcome (Figure 4B). A further cluster analysis identified two subtypes of THCA based on the core genes of the E2F-targets pathway (Figure 4C). Of the E2F-Clust, CS1 (cluster 1) was characterized by an evidently high expression of CAPN8 as well as core genes in E2F-targets pathways, showing an unfavorable effect on patients’ survival outcome (Figure 4D) compared to the superior influence of CS2 (cluster 2). This E2F-Cluster further supported that CAPN8 may lead to THCA progression by regulating the E2F-targets pathway.
There was a considerable diversity of biological function and TIME pattern between the two E2F-Clusters. CS1 was identified by the elevated pathway activity of E2F-targets, epithelial–mesenchymal transition (EMT), inflammatory response, and interferon-gamma response (Figure 5A). Compared to the two E2F-Clusters, three subtypes were identified in the ImmClust with a significant difference in survival outcome (Figure 5B). Considering the information of 10 immune cells and eight IRs, CS1 of the ImmuClust was accompanied with an increased expression of IRs and an infiltrating ratio of almost all types of immune cells, underlying a strongly inhibitory TIME pattern in the CS1 group. This result suggested that CAPN8 could induce T cell exhaustion to inhibit immune response and lead to a poor prognosis of THCA (Figure 5C). Moreover, CS1 was characterized by a higher infiltrating proportion of fibroblast than CS2, accounting for its distinctly higher enriched score of SES (Figure 5D). CS2 of the ImmClust, however, was characterized by an inflammatory TIME pattern with high levels of infiltration of neutrophil and endothelial cells. Distinctively, CS3 of the ImmClust lacked immune infiltration, suggesting a potential phenotype of immune desert for this subtype (Figure 5C). Keeping consistent with the exhausted TIME feature of CS1, immunotherapy seemed to be a feasible strategy for this subtype. Patients in CS1 significantly responded to PD1-R treatment (Figure 5E).
Oncoplot was illustrated to resolve the mutation profile of E2F-Clusters and ImmClusters. CS2 of the ImmClust seems to have the highest mutation frequency of BRAF and ZFHX3, while CS1 exceeded the other two clusters in the mutation frequency of COL5A3 and AKT1, which is a hub element of the PI3K proliferation pathway (Figure 6A). By contrast to the mutation frequency, CS2 of the ImmClust was fairly stable in CNV with the lowest FGA, FGL, FGG, and CN GISTIC score than CS1 and CS3 (Figure 6B), suggesting a rather recent cancer origin and fairly low chromosomal instability for this subtype. CS3, however, was in the lead in CNV frequency and CN GISTIC score, underlying a quite earlier cancer origin and extremely high chromosomal instability (Figure 6C) for this subtype.
With regards to the 21 pathways related to DNA replication stress, CS3, the immune desert subtype, was dramatically downregulated in the pathway activity of cell cycle, G1S-DNA damage checkpoints, G2M-DNA damage checkpoint, and mitotic spindle checkpoint, implying a considerably declined ability to maintain the correct paradigm of DNA replication (Figure 7A). This is in line with the result mentioned above, namely: there was a highly altered genomic CNV situation and increased chromosomal instability for CS3 cluster (Figure 6C), suggesting that CS3, the immune desert subtype, was prone to being drug resistant by inducing extensive epigenetic variations. As expected, CS2 of the E2F-Clust, similar to CS3 of the ImmClust, was relatively insensitive to many chemotherapeutics and correlated with poor survival outcome (Figure 7B). Specifically, the estimated IC50 values of VE-822, AZD67738, VE821, and MK-1775 were widely increased (Figure 7C). These drugs are famous inhibitors of ATR and week1, which are famous cell cycle regulatory proteins.
The z-score normalization for GSE3467 (n = 9), GSE3678 (n = 7), GSE33630 (n = 49), and GSE60542 (n = 33) was appropriate as the heterogeneity among the four datasets was eliminated after combination. After applying the E2F-Clust and ImmClust to this combined thyroid cancer cohort (n = 78), the same TIME pattern in the TCGA-THCA cohort was still recognizable. CS3 remained to be the phenotype of immune desert with decreased infiltration of immune cells and expression of IRs. CS1 remained as the phenotype of immune exhausted for its increased infiltration of immune cells and IRs expression, while CS2 maintained the phenotype of inflamed TIME with elevated infiltration of neutrophils (Figure 8A). The pattern of DNA replication stress, however, seems to be indistinguishable in the combined cohort. CS1, CS2, and CS3 were all seemingly accompanied by the increased activity of 21 DNA replication-related pathways (Figure 8A). Despite this uncertain result, CS2 was still much more likely to be drug resistant, with generally rising IC50 for ATR and week1 inhibitors (Figure 8C). Moreover, immunotherapy was plausible for the CS1 subtype in this combined cohort as patients in the CS1 subtype significantly responded to PD1-R treatment (Figure 8E). Similar patterns of TIME, drug sensitivity, and immunotherapy response were still distinguishable in another validation cohort: GSE27155 cohort (n = 95), annotated by GPL96 platform. The immune exhausted inflammatory and immune desert phenotypes still corresponded to CS1, CS2, and CS3 subtypes, respectively (Figure 8B). The activities of 21 DNA replication stress-related pathways and drug sensitivity kept decreasing in the CS3 subtype (Figure 8D). The CS1 group demonstrated the same certain probability of benefiting from PD1-R treatment (Figure 8F).
The present study identified three immune subtypes of THCA according to the expression of CAPN8 and related pathways, including the immune exhausted (CS1), inflamed (CS2), and immune desert (CS3) phenotypes. Three sub-clusters for THCA demonstrated quite diverse patterns of TIME, genetic variation, drug sensitivity, immunotherapy response, and patient prognosis—for instance, patients with CS1, with high expression of CAPN8, demonstrated rather detrimental survival outcomes when receiving chemotherapy but were much more likely to respond to anti-PD-1 treatment. These findings will provide novel insights for the treatment of THCA and help to understand the cross-talk between CAPN8 and the tumor immune microenvironment. Firstly, CAPN8 was found to be a significant risk factor for THCA with a markedly elevated level of mRNA and protein in tumor tissues. This potential oncogene also induced the activation of EMT and E2F-targets, which are well-acknowledged pathways to promote cancer metastasis and proliferation. Consistent with our studies, CAPN8 was claimed to be a potential oncogene in gastric cancer, hepatic carcinoma, and lung cancer, causing the occurrence of precancerous lesions and cancer progression (5, 6, 28). In addition, members of CAPN family have been reported to facilitate the invasion of THCA by inducing MMP2 and MMP9 secretion, which can contribute to extracellular matrix degradation during cancer cell migration (29). Furthermore, three immune phenotypes of THCA were identified according to the expression of CAPN8 and related pathways. CS1, the immune exhausted subtype, was accompanied with a distinctly increased expression of IRs and proportion of infiltrating T cells. Accordingly, exhausted T cells lose their killing ability because of the increased expression of IRs but can be restored by immune checkpoint inhibitors (30). This exactly accords with our result: CS1 was found to be positively responsive to anti-PD-1 treatment. In terms of CS3, this immune desert subtype of THCA demonstrated the absence of anti-tumor immunity and surging level of CNV. This is not surprising as there was a general downregulation of 21 pathways related to DNA replication stress in CS3. It is widely accepted that replication stress plays a key role in initiating anti-tumor immunity by inducing cancer-related neoantigens (31, 32), and its absence is intensively correlated with the immune desert phenotype. Consequently, the low immunity and high level of CNV can jointly contribute to strong cancer stemness (33), accounting for the result that patients of CS3 were tolerant of many ATR and week1 inhibitors in our study. Overall, three immune subtypes of THCA were identified based on the expression of CAPN8 and related pathways in our study. The three types displayed rather different paradigms of TIME, immune therapy response, drug sensitivity, and genomic variance. Moreover, these three immune subtypes are highly coincident with the results of previous studies on cancer classification (34–38), where the exhausted, inflamed, and desert phenotypes of breast cancer, prostate cancer, and bladder cancer were characterized by using a similar clustering analysis. Our study has several advantages. This is the first study to elucidate the role of CAPN8 in the metastasis of THCA from the aspects of TIME, DNA replication stress, and genetic alteration. Three immune subtypes identified in our study will provide new insights for the treatment of THCA, as different subtypes showed distinctly different responses to immunotherapy and chemotherapy. Most importantly, external validation in five individual cohorts made the extrapolation and robustness of the classification convincing. There were also some limitations to the current study. Firstly, further in vitro experiments will make it more authentic for the existence of the three subtypes of THCA. Secondly, the prognostic effects of CAPN8 could be validated in actual cohorts of THCA to make it more persuasive. Lastly, analysis on cancer stemness can be added to further explain the relationship between different subtypes and various drug sensitivities. In conclusion, we highlighted the role of CAPN8 in THCA metastasis and identified three distinct immune subtypes that can be distinguished in terms of prognosis, immunotherapeutic response, and drug sensitivity, which provide new insights for the treatment of THCA and contribute to the understanding of the interaction between CAPN8 and the tumor immune microenvironment.
The original contributions presented in the study are included in the article/Supplementary Materials. Further inquiries can be directed to the corresponding authors.
All authors contributed to the study’s conception and design. XZ, SX, and QW performed data collection and analysis. XZ and SX wrote the manuscript. LP, FW, and TH polished and revised the manuscript. ZH and SN provided analytical ideas. All authors contributed to the article and approved the submitted version.
This work was supported by grants from the Research Project of Maternal and Child Health of Jiangsu Province (F201953) and the Science and Technology Project of Nantong (JC2020067) to ZH.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647054 | Alessio Stefani,Geny Piro,Francesco Schietroma,Alessandro Strusi,Emanuele Vita,Simone Fiorani,Diletta Barone,Federico Monaca,Ileana Sparagna,Giustina Valente,Miriam Grazia Ferrara,Ettore D’Argento,Mariantonietta Di Salvatore,Carmine Carbone,Giampaolo Tortora,Emilio Bria | Unweaving the mitotic spindle: A focus on Aurora kinase inhibitors in lung cancer | 27-10-2022 | lung,cancer,oncology,Aurora kinase (AURK),mitosis,AURK inhibitors | Lung cancer is one of the most aggressive malignancies, classified into two major histological subtypes: non-small cell lung cancer (NSCLC), that accounts for about 85% of new diagnosis, and small cell lung cancer (SCLC), the other 15%. In the case of NSCLC, comprehensive genome sequencing has allowed the identification of an increasing number of actionable targets, which have become the cornerstone of treatment in the advanced setting. On the other hand, the concept of oncogene-addiction is lacking in SCLC, and the only innovation of the last 30 years has been the introduction of immune checkpoint inhibitors in extensive stage disease. Dysregulation of cell cycle is a fundamental step in carcinogenesis, and Aurora kinases (AURKs) are a family of serine/threonine kinases that play a crucial role in the correct advance through the steps of the cycle. Hyperexpression of Aurora kinases is a common protumorigenic pathway in many cancer types, including NSCLC and SCLC; in addition, different mechanisms of resistance to anticancer drugs rely on AURK expression. Hence, small molecule inhibitors of AURKs have been developed in recent years and tested in several malignancies, with different results. The aim of this review is to analyze the current evidences of AURK inhibition in lung cancer, starting from preclinical rationale to finish with clinical trials available up to now. | Unweaving the mitotic spindle: A focus on Aurora kinase inhibitors in lung cancer
Lung cancer is one of the most aggressive malignancies, classified into two major histological subtypes: non-small cell lung cancer (NSCLC), that accounts for about 85% of new diagnosis, and small cell lung cancer (SCLC), the other 15%. In the case of NSCLC, comprehensive genome sequencing has allowed the identification of an increasing number of actionable targets, which have become the cornerstone of treatment in the advanced setting. On the other hand, the concept of oncogene-addiction is lacking in SCLC, and the only innovation of the last 30 years has been the introduction of immune checkpoint inhibitors in extensive stage disease. Dysregulation of cell cycle is a fundamental step in carcinogenesis, and Aurora kinases (AURKs) are a family of serine/threonine kinases that play a crucial role in the correct advance through the steps of the cycle. Hyperexpression of Aurora kinases is a common protumorigenic pathway in many cancer types, including NSCLC and SCLC; in addition, different mechanisms of resistance to anticancer drugs rely on AURK expression. Hence, small molecule inhibitors of AURKs have been developed in recent years and tested in several malignancies, with different results. The aim of this review is to analyze the current evidences of AURK inhibition in lung cancer, starting from preclinical rationale to finish with clinical trials available up to now.
Despite the continuous progress in understanding its biology and discovering new potential targets, lung cancer is responsible for the highest number of cancer-related deaths in Italy (1). Non-small cell lung cancer (NSCLC) represents about 85% of lung cancer new diagnoses and it is a heterogeneous disease, often characterized by the presence of a driver mutation (oncogene-addicted disease) for which a targeted drug is available. The introduction of immune checkpoint inhibitors (ICIs) has changed the history of non-oncogene addicted disease: immunotherapy, alone or in combination with chemotherapy, represents the standard first-line treatment, reaching the biggest benefit in patients with strong expression of PD-L1 (5 years OS: 31.9% vs 16.3% with platinum-based chemotherapy) (2). Small-cell lung cancer (SCLC) represents the other 15% of lung cancer diagnoses; it is an aggressive disease, with a high proliferation rate and a high dissemination potential, in fact most cases are diagnosed at an advanced stage. Genomic profiling of SCLC identified p53 and pRB as the most frequently altered genes (3), but no targeted therapies are available up to now. Therefore, SCLC is treated as a single entity and platinum-based chemotherapy has been considered the standard of care for the last thirty years. Since the results of IMpower133 and CASPIAN trials, immunotherapy in combination with platinum-etoposide has become the new recommended first-line treatment; although the global benefit of ICIs is small (ΔmOS=2 months), about 15-18% of patients experience a long-term benefit, being alive at 18 months after treatment start (4, 5). Due to the limited options available after the failure of first-line regimens, particularly in SCLC, research efforts must focus on expanding the therapeutic strategies for lung cancer. An increasing attention has been focused on cell cycle regulators targeting drugs. One of the main actors in cell cycle are Aurora kinases (6, 7). Their importance was initially highlighted by genetic studies on mutants demonstrating their role in the abnormal mitotic spindle formation (from which the name “aurora”, resembling aurora borealis) and cytokinesis failure. In this review, we will focus on the rationale of targeting Aurora kinases in lung cancer, disclosing the results of the available clinical trials.
Aurora kinases (AURKs) are a family of serine/threonine kinases that plays fundamental roles in cell cycle, particularly in mitotic spindle formation and in chromosome segregation. In mammals, there are three known members of this family: Aurora kinase A (AURKA), Aurora kinase B (AURKB) and Aurora kinase C (AURKC). AURKs are composed of three domains: a N-terminal domain the kinase domain and a C-terminal domain. The catalytic domain shares >70% of homology among the three isoforms (8) and is composed of a β-stranded lobe and an α-helical lobe, linked by a hinge region; the two lobes create a deep cleft where the ATP-binding pocket lies (9). The non-catalytic domains are likewise essential for the correct functions of AURKs: the N-terminal domain mediates the intracellular localization, while the C-terminal domain binds to specific co-factors that shape their conformation (10). The kinase action is only activated after auto-phosphorylation of a specific threonine residue in the catalytic domain. The specific roles of Aurora kinases depend on the different intracellular localization and the meticulous temporal expression during the cellular cycle. Transcription of AURKs is regulated by cell cycle-dependent factors that bind to cell cycle-dependent elements (CDE) in their promoters (11). AURKC seems to be significantly expressed only in cells undergoing meiosis (i.e., spermatocytes and oocytes) and its biological functions are not well understood. Although it is overexpressed in many cancer types (12), its oncogenic role is unclear; however, it may be responsible for centrosome amplification and multinucleation of cancer cells, conferring survival advantage (13). AURKA and AURKB are, on the contrary, expressed in every cell undergoing mitosis. AURKA levels rise from G2 phase to early mitotic phases (14–16); at first, AURKA can be found in the pericentriolar matrix and, after activation by co-factor Ajuba, it contributes to centrosome maturation: AURKA recruits several proteins essential to microtubule nucleation, stabilization and spindle assembly, like centrosomin, γ-tubulin ring complex (γ-TuRC) and D-TACC/maskin (17, 18). During late prophase, AURKA phosphorylates cyclin B1-Cyclin-Dependent Kinase 1 (CDK1), which, in turn, provokes the nuclear envelope breakdown (NEBD) by activating the Ran GTPase pathway. After NEBD, AURKA is responsible for centrosome separation by phosphorylating kinesin Eg5, which generates a sliding movement on anti-parallel microtubules pushing the centrosomes away (19). Cyclin B1-CDK1 complex also activates the spindle assembly factor TPX2, which binds to AURKA and, together, they create the bipolar mitotic spindle (20–22). During early mitosis, AURKB phosphorylates histone H3 in order to release heterochromatin protein 1 (HP-1) from heterochromatin; this event might facilitate chromosome condensation, but evidence is unclear in mammalian cells (23, 24). Then, during prophase, AURKB regulates the attachment of microtubules of mitotic spindle to kinetochores. Kinetochores are protein complexes that bind to chromatin domains which act as a platform called centromeres. AURKB is a member of the error correction (ER) machinery, a control system that detects tension between centromere and kinetochore and stabilizes correct chromosome biorientation (amphitelic), whereas it inhibits incorrect “tensionless” attachments (such as synthelic, monothelic and merotelic) (25). Furthermore, in case of incorrect attachments, AURKB activates the spindle assembly checkpoint (SAC) that prevents sister chromatids separation and mitotic exit (26, 27). During metaphase, AURKB takes part of the chromosome passenger complex (CPC), together with INCENP (inner centromeric protein), Survivin and Borealin, and relocates to the midzone (28). It has been shown in yeasts that AURKB promotes sister chromatid separation by recruiting Shugoshin 1 (SGO1), that removes Cohesin from centromeres (29). Lastly, AURKB plays an essential role in cytokinesis: the activation of RhoA GTPase determines actine polymerization and the formation of the contractile ring; phosphorylation of vimentin, desmin and GFAP creates the cleavage furrow (30). Given the crucial roles in cell cycle, activity of Aurora kinases must be finely regulated, particularly in case of DNA damages. When G2 checkpoint is activated by double strand breaks, ATM and ATR phosphorylate checkpoint kinase Chk1/Chk2, that not only inhibits cyclin B1, but also AURKA and AURKB; AURKB is also blocked by PARP1 (31, 32).
Dysregulation of Aurora kinases can lead to proliferative and survival advantages in many tumors. Although there are no validated methods to assess AURK overexpression, different techniques could be used including immunohistochemistry, FISH and comparative multiplex RT-PCR, that can detect differential AURK-mRNA expression in normal and tumor tissues. Overexpression of AURKA is found in different cancers, including lung carcinomas, and is an established poor prognostic factor in lung, breast and colorectal cancers (33–35). The induction of AURKA overexpression in vitro did not demonstrate the capacity of transforming cell lines or generating malignant tumors in murine models, so Aurora A might rather be a promoting factor than an oncogene (36). In fact, AURKA overexpressing cells are characterized by multipolar spindle formation and unequal chromosome segregation, leading to aneuploidy and a potentially precancerous state. Moreover, abnormal AURKA activity hyperactivates oncogenic pathways like NFκβ, BCR/ABL and Pi3K/Akt, resulting in increased cell proliferation, survival and transformation. AURKA is also able to upregulate telomerase activity via hyperactivation of Myc, leading to increased survival (37). Lastly, AURKA is linked to epithelial-to-mesenchymal transition (EMT) and metastatic potential in several cancer (38, 39). Yoo and colleagues recently showed that AURKA and AURKB confer an “invasiveness signature” in lung adenocarcinoma, indeed their simultaneous inhibition in vitro and in a murine model of lung adenocarcinoma reduced tumor invasion (40). AURKB is found overexpressed in many cancer types (41, 42) and is a negative prognostic factor in NSCLC and hepatocellular carcinoma amongst other tumors (43, 44). Abnormal expression of AURKB is linked to aneuploidy and micronuclei formation, in fact its overexpression alters chromosome segregation and SAC activation (45); in p53-deficient cells, these effects are even augmented (46, 47). AURKs dysregulation is also responsible for resistance to several antineoplastic drugs. In a recent study by Tagal and colleagues, it was shown that AURKs could determine a switch from the proliferative cell cycle to polyploid growth and multinucleation in lung cancer cell lines, resulting in the formation of polyploid giant cancer cells (PGCC) (48). These cells seem to be associated with resistance to many antimitotic drugs, tumor relapse, immunosuppression, cancer stem cell production, and modulation of the tumor microenvironment (49). Expression of aurora A kinase is correlated with cisplatin resistance in NSCLC: in vitro data of 102 NSCLC patients treated with surgery and adjuvant cisplatin-based chemotherapy showed that AURKA expression was elevated in cisplatin-resistant lung cancer cells. Furthermore, its inhibition reversed the migration ability of cisplatin-resistant cells (50). High levels of AURKA are also associated with cisplatin resistance in JAK2-mutated myeloma cells (51). AURKB’s expression modulates the activity of taxanes in NSCLC cells and the assessment of its levels in histological samples could be developed as a predictive biomarker. It has been shown that mRNA expression of AURKB in NSCLC cell lines inversely correlated with resistance to both docetaxel (p = 0.004) and paclitaxel (p = 0.007). Furthermore, inhibition of AURKB activity with barasertib also demonstrated a strong dose-dependent efficiency in triggering paclitaxel resistance. The results of the study bring to a paradox: overexpression of AURKB reduces survival in chemotherapy-naive patients but, on the other hand, it appears to have a beneficial effect in patients treated with taxane regimens (52).
In a large cohort of NSCLC patients (n = 362) AURKA was highly overexpressed in the tumor tissues compared to corresponding normal lung tissue. In univariate analyses it resulted a significantly increased hazard ratio and poor disease-free survival in patients with a high gene expression of both AURKA (HR = 2.813, p ≤ 0.001) and its co-factor TPX2 (HR = 1.826, p = 0.007). Similarly, AURKA expression confirmed to be a statistically significant prognostic marker using multivariate analyses (p = 0.006) (35). A study including 11 NSCLC cell lines investigated the preclinical efficacy of MK-5108, a strong inhibitor of AURKA that had shown a potent preclinical activity in malignancies of breast, cervical, colon, ovarian, and pancreatic origin (53). MK-5108 was tested as a single agent and in combination with cisplatin and docetaxel. Concurrent treatment of MK-5108 with cisplatin or docetaxel synergistically inhibited cell growth, with the docetaxel combination performing better. In sequential administration, treatment with docetaxel followed by MK-5108 registered greater growth inhibition than the inverse, even if concurrent treatment remained superior (54). Different preclinical studies focused on the role of AURKs in oncogene-addicted NSCLC and in particular on their role in the induction of resistance to targeted therapies. Activating mutations in the Epidermal Growth Factor Receptor (EGFR) gene are the most frequent mutations and they can be found in 14–17% of advanced NSCLC in European populations (55). Tumors with common mutations are sensitive to EGFR tyrosine kinase inhibitors (EGFR TKIs), but eventually these patients will develop resistance which will lead to disease progression. Treatment-induced activation of AURKA seems to be associated with in vitro and in vivo resistance to EGFR inhibitors. In response to chronic EGFR inhibition, AURKA can be activated by the overexpression of TPX2, which facilitate its auto-phosphorylation; TPX2 is normally degraded by a ubiquitin E3 ligase, which is intra-nuclear in both parental and resistant cells (56). In contrast, in resistant cells TPX2 delocalize in the cytosol, separate from the complex responsible for its degradation, leading to its accumulation. Aurora kinase inhibitors suppress this adaptive survival program, increasing the magnitude and duration of EGFR inhibitor response in preclinical models. The suppression of AURKA-driven residual disease could become an important weapon against the acquired resistance in these diseases. The combination of an aurora kinase inhibitor with a third-generation anti-EGFR agent resulted in a synergistic reduction in cell growth in all models (57). In addition, AURKA overexpression is linked to acquired resistance to EGFR-TKI via epithelial-mesenchymal transition (EMT), and AURKA inhibitor alisertib has shown to restore NSCLC cells sensitivity to EGFR-TKI and to partially reverse EMT (58). AURKA inhibition with shRNA also demonstrated to partially reverse fibroblast-mediated resistance to gefitinib in EGFR-mutated NSCLC cells co-cultured with stromal cells (59). Another study showed that resistant EGFR-mutated NSCLC cells without the p.T790M or other acquired mutations are sensitive to AURKB inhibitors barasertib and S49076. In most acquired resistant cells in fact the phospho-histone H3 (pH3), a major product of AURKB, resulted increased and its levels reduced after treatment with AURKB inhibitors, triggering G1/S arrest, polyploidy and, eventually, cell cycle arrest and cell death. The results support the role of AURKB activation in acquired resistance to EGFR TKIs, making AURKB a potential target in NSCLC progressed to anti-EGFR therapy and not carrying resistance mutations (60). AURKB inhibitors are potent enhancers of osimertinib-induced apoptosis and can play an important role in overwhelming acquired resistance to third generation TKIs. Osimertinib resistance caused by EMT activates the ATR-CHK1-Aurora B signaling cascade and generates hypersensitivity to AURKB inhibitors by activating BIM-mediated mitotic catastrophe. AURKB inhibition stabilizes BIM through reduced Ser87 phosphorylation, and transactivates PUMA through FOXO1/3. In this way a combined inhibition of EGFR and AURKB not only efficiently eliminates cancer cells but also overcomes resistance beyond EMT (61). AURKA and B have also shown to phosphorylate KRAS downstream effectors, playing a synergic oncogenic role with KRAS mutations. Dos Santos et al. demonstrated that KRAS positively modulated AURKA and AURKB expression by regulating their transcription or mRNA stability. They also assessed that simultaneous pharmacological inhibition of AURKA and AURKB activity in vitro, as well as their targeting by RNA interference, reduced cell growth and proliferation and promoted apoptosis in a KRAS-dependent manner. Unfortunately, these results were not confirmed in in vivo xenografts model; however, this study suggests that aurora kinases could be targeted in KRAS-mutated NSCLC (62). According to results presented at the IASLC 2022 World Conference on Lung Cancer, Lee et al. demonstrated that the addition of AURKA inhibitor VIC-1911 to KRAS inhibitor sotorasib led to increased cell death in resistant cancer cells compared to sensitive ones, suggesting that AURKA inhibition may overcome sotorasib resistance. In addition, the combined inhibition of AURKA and WEE1 led to a synergistic increase in the death of KRAS-mutated lung cancer cells with acquired resistance to sotorasib, even greater than sotorasib plus VIC-1911 (63). AURK inhibitors were investigated as radiosensitizing agents by Liu et al. in NSCLC cell lines. MLN8237 (alisertib) was assessed together with the effect of radiation and, after treatment, p53-proficient HCC2429 and H460 cell lines increased their sensitivity to the lethal effect of radiation, with a dose enhancement ratio (DER) of 1.33 (p < 0.05) and 1.35 (p < 0.05), respectively; on the other hand, there was no significant enhanced effect in the naturally p53-deficient and radiation-resistant H1299 cells with a DER of 1.02 (p > 0.05). These data suggest that lower doses of radiation could achieve an equivalent antitumor effect when administered in combination with MLN8237 compared to radiation alone in vitro, especially in p53-competent cells (64). Taking into account these early signs of preclinical activity, the role of AURK inhibitors in NSCLC has also been investigated in clinical trials (synthetized in Table 1 ). A multicenter, 5-arm, phase II trial investigated the safety and activity of single-agent alisertib in various advanced and pretreated solid tumors (n = 249). Alisertib was administered orally in 21-day cycles at the recommended dose of 50 mg twice daily for 7 days followed by a break of 14 days. The study included 26 patients with NSCLC and an objective response (OR) was registered in just 1 (4%, 0-22) of 23 evaluable patients, while 17 (74%, 52-90) achieved a stable disease (SD). In the NSCLC cohort, 25 patients (96%) experienced an adverse event (AE) of any grade and the most frequent drug-related grade 3-4 adverse events included neutropenia (62%), leukopenia (27%), fatigue and anemia (both 19%). Despite the manageable toxicity profile, the activity data of alisertib were not particularly promising in patients with NSCLC and did not support further clinical assessment in this disease, in contrast to breast cancer and SCLC (65). Godwin and colleagues assessed whether the combination of erlotinib and alisertib exerted a synergistic action in EGFR wild-type NSCLC in a phase I/II clinical trial. 18 patients with recurrent or metastatic EGFR wild-type NSCLC were treated and the combination of alisertib and erlotinib proved to be tolerable. Common drug-related adverse events of any grade were fatigue (89%), anemia (83%), leukopenia (78%), dyspnea (78%), diarrhea and anorexia (61%), while drug-related grade 3/4 adverse events included neutropenia and leukopenia (33%), febrile neutropenia, lymphopenia, and anemia (11%). The maximum tolerated dose (MTD) was 150 mg daily for erlotinib with 40 mg BID for alisertib. Disease responses were also noted, including one patient with a partial response who completed 10 cycles, and 5 patients who achieved SD (66). A single-center phase I study including 17 patients with refractory advanced solid tumors investigated the safety and tolerability of alisertib combined with weekly irinotecan (100 mg/m2 on day 1 and 8 of a 21-day cycle). Alisertib was administered orally twice per day on days 1-3 and 8-10 with an escalating dose of 20-60 mg. The MTD was 20 mg twice per day and the dose-limiting toxicities were diarrhea, dehydration, and neutropenia. Furthermore, it was registered one fatal cardiac arrest at the highest dose level tested which was possibly related to drug. No objective responses were observed in patients with NSCLC. Due to the weak activity and most of all to the poor tolerance, the use of alisertib in combination with irinotecan did not show appealing results (67). Blackely et al. presented at the 2021 ASCO Annual Meeting the promising preliminary results of intermittent dosing of alisertib (30 mg BID on days 1-3, 8-11, and 15-17 of a 28-day cycle) in combination with osimertinib (80 mg daily) in patients with EGFR-mutated lung adenocarcinoma resistant to osimertinib monotherapy. In this phase Ia clinical trial (NCT04085315) 6 patients were treated with 30 mg BID and 4 patients with 40 mg BID intermittent dosing schedule of alisertib. The most commonly reported adverse events were diarrhea (70%), fatigue (60%), alopecia (50%) and neutropenia (50%), all of them of grade 1 or 2; two patients (20%) experienced grade 3 or grade 4 neutropenia, both patients were treated at the 40 mg BID intermittent dose of alisertib. Intermittent alisertib 30 mg BID was identified as the MTD and recommended phase 2 dose in combination with osimertinib 80 mg daily. The ORR was 10% (1/10) and DCR 70% (7/10). The median PFS was 9.4 months (2.0 months - N.R.) (68). AT9283, an inhibitor of AURKA and AURKB, has been assessed in a phase I dose-escalation study in 49 patients with advanced solid tumors including NSCLC (n = 7). This drug was generally well tolerated with reversible dose-related toxic effects such as myelosuppression, gastrointestinal disturbance, fatigue, and alopecia. No objective responses were observed; however, four patients with esophageal cancer (n = 1), colorectal cancer (n = 1), and NSCLC (n = 2) demonstrated prolonged SD of more than 6 months (69). The role of another AURKB inhibitor (BI 811283) was investigated in a phase I, dose-escalation study involving 121 patients with advanced solid tumors. The drug was administrated via 24-hours infusion on Days 1 and 15 of a 4-week cycle (schedule A) or Day 1 of a 3-week cycle (schedule B) and the MTDs obtained were 125 mg and 230 mg respectively. 4 patients with NSCLC were included in this study: 3 were treated with schedule A and 1 with schedule B. All patients in both treatment schedules experienced at least one adverse event. The most common dose-limiting toxicities were hematological events, particularly neutropenia. Pharmacodynamic assessments showed a decrease in phosphorylated histone H3 (pHH3) which indicated Aurora B kinase inhibition. No patient achieved an OR, even if 30% in schedule A and 33% in schedule B reported a clinical benefit and a stabilization of the disease. Despite a good safety profile, the anti-tumor activity observed does not support the development of the drug in solid tumors (70). In a prospective, phase II, open-label, multi-institutional study, Danusertib (PHA-739358, a pan-AURK inhibitor) was adopted as single agent for treating patients with different advanced cancers including NSCLC as second line treatment. Patients were treated with danusertib 500 mg/m2 given as 24-h i.v. infusion every 14 days until progression or unacceptable toxicity. Danusertib showed marginal antitumor activity with a manageable safety profile. In the 56 patients with metastatic NSCLC the progression-free rate (PFR, the primary outcome) at 4 months was 10.4% (16.1% in squamous subgroup, where the only objective RECIST response was obtained). The mPFS was 9.2 weeks and the mOS 7.6 months. AEs were reported in 83.3% of patients. The most frequent drug-related AEs were fatigue (67.9%), nausea (39.3%), diarrhea (28.6%), anorexia (28.6%), vomiting (16.1%), alopecia (23.2%), constipation (10.7%), anemia and neutropenia (74.5% of events CTC grade 3 or 4) (71). Barasertib (AZD1152), another Aurora kinases inhibitor, was tested in two phase I studies. Patients with different advanced solid malignancies were treated with escalating doses (100-650 mg) administered as a 2-h infusion every 7 days or 14 days. The MTD was respectively 200 mg and 450 mg, and neutropenia was the most frequent adverse event and dose-limiting toxicity. Grade 3-4 neutropenia occurred in 58% and 43% of patients. No OR were observed at any dose or schedule, although 15 patients (25%) achieved a SD. However, only 3 patients had NSCLC and that is why the role of barasertib is far from being defined in this type of tumor (72, 73).
Even after the introduction of immunotherapy in the first-line setting, the majority of patients with SCLC experiences an inexorable disease progression in less than 12 months (4, 5). Unfortunately, effective treatments are not available after disease progression to first-line therapy: topotecan is currently the standard of care, with limited results (74). These poor outcomes highlight the need for a better molecular knowledge of the disease to develop new therapeutic strategies. The most common genetic mutations of SCLC are related to p53 and RB1, but none of these represent a druggable therapeutic target. Amplification of MYC family genes was also found in about 20% of SCLCs (75) and in 30-50% of SCLC cell lines (76) and is associated with treatment resistance, tumor progression and poor outcomes (77, 78). Recent studies have shown that the SCLCs family can be divided into four distinct subtypes based on the differential expression of four transcription factors (79); two of these subgroups, characterized by a high expression of ASCL-1 (SCLC-A) or NEUROD1 (SCLC-N), share a neuroendocrine phenotype; the other two subgroups can be divided on the basis of the expression of POU2F3 (SCLC-P) or of the lack of expression of the three transcription factors (SCLC-I). This last subgroup is instead characterized by the expression of an immunogenic signature, including immune checkpoints and human leukocyte antigens (HLAs), therefore the denomination “inflamed” (80). In a study conducted on murine models, SCLC-N appeared to be associated with MYC amplifications (81, 82). In fact, data suggest that MYC promotes a variant subset of SCLC with lower expression of neuroendocrine markers and with more aggressive features, that could originate from ASCL1+ progenitor cells which, over time, transition to an ASCL1-low/NEUROD1-high state due to the indirect effect of MYC on NEUROD1 signaling (83). Despite these findings, it is still difficult to exploit MYC in a therapeutic way. Nevertheless, from synthetic lethality screenings, AURK inhibitors appeared promising candidate targets. Mollaoglu et al. demonstrated that MYC-driven SCLC cell lines were sensible to AURKA inhibitor Alisertib and AURKB inhibitor Barasertib. To assess AURK inhibition in vivo, murine models bearing MYC-amplified SCLC received Alisertib alone, chemotherapy alone or chemotherapy + Alisertib. While single agent alisertib or chemotherapy didn’t show durable results, mice who received the combination had the highest 30-day survival rate (47% vs 5% for chemo-treated vs 8% for Alisertib-treated) (83). Antitumor activity in vivo of these molecules was tested in few clinical trials (synthetized in Table 2 ). A phase I dose-escalation trial tested Danusertib as a 24-hour infusion with and without G-CSF in patients with advanced pretreated solid tumors. Among the 56 patients enrolled in the study, 2 had SCLC. One of these patients experienced an objective tumor response that lasted for 23 weeks receiving 1,000 mg/m2 Danusertib + G-CSF, subsequently reduced to 750 mg/m2 for hypercreatininemia G2. Drug related SAEs occurred in 21% of all patients (12/56), 9 (22%) in the group treated with Danusertib alone and 3 (19%) in the group treated with Danusertib + G-CSF (84). A subsequent multi-cohort phase II study included 18 patients with SCLC who had failed at least two prior lines of therapy that were treated with Danusertib (multi AURK-inhibitor). Unfortunately, none of these patients was progression-free at the four-month treatment assessment. Final results have shown a mPFS of 8.11 weeks and a mOS of 11.4 months. Regarding its safety profile, Danusertib confirmed what had already emerged from previous studies: the most frequent treatment-related non-hematological AEs were asthenia/fatigue (61%, 11/18) and nausea (38.9%, 7/18); neutropenia was the most common hematological toxicity (100%) as well as the most frequent grade 3–4 event (88.9%, 16/18) (71). A five-arm phase II study investigated the activity of Alisertib in 60 patients with pretreated SCLC. Results have shown that, among response-assessable patients, an OR was obtained in 21% (10/48). The most frequent drug-related grade 3–4 adverse events included neutropenia, leukopenia, and anemia (65). Lastly, a phase I trial studied AMG 900, an orally administered pan-Aurora Kinase inhibitor in patients with advanced solid tumor. Among the 105 patients treated in this trial, 3 patients of the escalation cohort had SCLC. Unfortunately, none of them obtained an OR with the treatment. Regarding the safety profile, treatment-related AE with grade ≥ 3 occurred in 61 patients (58%); the most common one was neutropenia (n=44, 42%). The most common non hematological AEs were fatigue and diarrhea (85). The activity and safety of the association of chemotherapy with aurora kinase inhibitors was evaluated in a few clinical trials (synthetized in Table 3 ). In the previously reported phase I study investigating the combination of alisertib and irinotecan in solid tumors, 3 of 17 patients had a diagnosis of SCLC. Although one PR occurred in a patient with SCLC among the 11 evaluable patients (9%), the toxicity profile showed significant rates of toxicities hematological and gastrointestinal toxicities, leading the authors to conclude that the combination of Alisertib and Irinotecan was not well tolerated in adult patients and to stop the planned expansion cohort (67). Another phase I trial in patients with advanced solid tumors tested the combination of Alisertib and nab-paclitaxel, with the rationale of combining their antimitotic action. Among the 31 patients treated in the dose-escalation phase, 5 had a diagnosis of SCLC. Results have shown that one patient with refractory SCLC achieved a partial response that lasted for more than two years, until treatment was discontinued due to neurological toxicities. Two other patients with SCLC achieved a SD that lasted more than four months. These data led to an OR of 6.3% (1/16) and a DCR of 31.3% (5/16) among the 16 evaluable patients. Regarding the safety profile, the most common treatment-related AEs included alopecia (64.5%), diarrhea (41.9%), oral mucositis (41.9%), anorexia (38.7%), fatigue (38.7%), and nausea (35.5%). The most common laboratory abnormalities were leukopenia (80.6%), neutropenia (77.4%) and anemia (77.4%) (86). A randomized double-blind phase II study assessed paclitaxel + alisertib/placebo as a second line treatment after platinum-based chemotherapy in 178 patients with SCLC, stratified by relapse type (sensitive vs refractory/resistant); mPFS was 3.32 months in the Alisertib + Paclitaxel arm versus 2.17 months in the Placebo + Paclitaxel arm (p=0.113), while mOS was 6.86 months versus 5.58 months (p=0.714). The DCR was 58% in the experimental arm versus 46% in the control arm, and ORR was 22% and 18% respectively. Slightly better results were shown in the subgroup of resistant/refractory patients. In addition, C-Myc-positive patients and those with mutations in genes involved in cell cycle regulation (CDK6, RBL1, RBL2, RB1) also showed better outcomes with Alisertib than with Placebo. The incidence of grade 3 or higher drug-related AEs was 67% with Alisertib + Paclitaxel versus 25% with Placebo + Paclitaxel; the most common AEs were neutropenia, febrile neutropenia, leukopenia, anemia, diarrhea and stomatitis (87). The combination of Alisertib + Docetaxel was evaluated in a phase I clinical trial in the context of solid tumors eligible for Docetaxel therapy as determined by the investigator. Among the 41 patients that participated, only one patient had a diagnosis of SCLC and did not achieve an objective response. Treatment-related grade 3 or higher AEs involved 39 patients (95%), and the most common one was neutropenia (n=34, 83%) (88). Lastly, it is worth reporting the case of a nonsmoker patient with SCLC harboring a novel JAZF1-MYCL1 gene fusion and lacking alterations in TP53 and RB1. The patient had previously been treated with chemo-radiotherapy in the setting of limited stage disease; subsequently, after disease recurrence, the patient was enrolled in a clinical trial with Alisertib as his fourth-line regimen and achieved an almost complete response after ten cycles; the patient discontinued treatment after approximately 18 months of therapy (23 cycles) due to disease progression, and after the failure of subsequent chemotherapy lines, obtained an excellent disease control with Nivolumab (89).
The role of Aurora kinases in regulating cell cycle and safeguarding the correct transmission of genome to daughter cells is well established. Dysregulation of AURKs showed to promote tumorigenesis with different mechanisms, particularly causing aneuploidy and favoring genome instability. In addition, overexpression of AURKs is related to antineoplastic drug resistance, particularly platinum compounds and EGFR-TKIs in the case of lung cancer. Despite the strong rationale in the use of AURK inhibitors against cancer, significant clinical activity was demonstrated in hematological malignancies (90–93) but not in many solid tumors; this different outcome might be explained by the higher proliferation rate and clonality of the formers. In NSCLC, AURK inhibitors showed weak antitumor activity; nevertheless, preclinical studies and early data from clinical studies support their investigation in combination with EGFR-inhibitors. An ongoing clinical trial is evaluating safety and activity of the combination of osimertinib + alisertib or sapanisertib (an oral inhibitor of TOR complex 1 and 2) in osimertinib-resistant EGFR-mutated lung cancer (NCT04479306). Similarly, another clinical trial will study AURKA inhibitor LY3295668 in combination with osimertinib in patients with advanced EGFR-mutant NSCLC who have received a third generation EGFR-TKI (NCT05017025). In SCLC, early-phase clinical trials showed appreciable signals of activity of AURK inhibitors, particularly in combination with taxanes, but these results need to be validated in phase III randomized trials. In addition, considering the better outcomes obtained in cMyc-positive tumors, efforts should be made to apply the concept of precision medicine even in SCLC; the four subgroups based on differential expression of transcription factors ASCL1, NEUROD1, POU2F3 and YAP1 could provide a reproducible method of classifying SCLC for this scope, considering that cMyc tends to be overexpressed in SCLC-N subtype.
ASte and GP did write the review, FS, AStr, EV, SF, DB, FM, IS GV, and MF did review all the specific literature and pooled all available data from peer-reviewed journal and featured oncology meetings, ED’A, MDS, and CC did critically review all the drafts and GT and EB did coordinate the whole work. All authors contributed to the article and approved the submitted version.
EB is supported by Institutional funds of Università Cattolica del Sacro Cuore (UCSC-projects D1) and by the Fondazione AIRC (Associazione Italiana Ricerca sul Cancro) under Investigator Grant (IG) No. IG20583.
EB received advisory and speakers’ fee from MSD, Astra-Zeneca, Celgene, Pfizer, Helsinn, Eli-Lilly, BMS, Novartis, and Roche. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647055 | Watchara Sakares,Wannaporn Wongkhattiya,Ponlawat Vichayachaipat,Chompunoot Chaiwut,Varalee Yodsurang,Pattiya Nutthachote | Accuracy of CCL20 expression level as a liquid biopsy-based diagnostic biomarker for ovarian carcinoma | 27-10-2022 | ovarian neoplasms,chemokine CCL2,chemokine CCL15,chemokine CCL20,chemokine CXCL14,biomarker | Objective The study aimed to investigate the potentiality of chemokines, including MCP-1, CCL15, CCL20, and CXCL14, as biomarkers for differential diagnosis between benign tumors and ovarian cancer (OC). Methods A cross-sectional study was conducted in women aged >18 years who had adnexal masses treated with elective surgery at the HRH Maha Chakri Sirindhorn Medical Center, Srinakharinwirot University, between 2020 and 2021. The preoperative MCP-1, CCL15, CCL20, and CXCL14 serum levels were measured using a sandwich enzyme-linked immunosorbent assay. Preoperative diagnosis was defined according to the risk of malignancy index. The histological diagnosis and cancer subtype were confirmed using pathological specimens. Results Ninety-eight participants were preoperatively diagnosed with malignant tumors. The pathological diagnosis confirmed OC in 33 patients and disclosed 27 misdiagnosed cases, of which endometriotic cyst was the most common (44.44%). CCL20 and CA125 serum levels were significantly higher in the patients with cancer than in those with benign. In addition, CCL20 level could differentiate between benign and early-stage malignancy. Furthermore, only CCL20 levels could distinguish endometriotic cysts from OC, whereas CA125 levels could not. Concordant with the serum protein level, the increased mRNA level of CCL20 was observed in ovarian cancers comparing with that in benign tissues. We found that CCL20 levels could differentiate between benign tumors and OC with 60.61% sensitivity and 75.44% specificity at the optimal cutoff value of 38.79 pg/ml. Finally, the logistic regression model integrating CCL20, CA125, and menopause status promoted diagnostic accuracy by increasing the specificity to 91.23%. Conclusions Our study revealed the potential usefulness of CCL20 level as a biomarker for diagnosing early-stage OC with endometriosis differentiation. We recommend further studies to confirm the accuracy of CCL20 levels with the current diagnosis in a large patient sample. | Accuracy of CCL20 expression level as a liquid biopsy-based diagnostic biomarker for ovarian carcinoma
The study aimed to investigate the potentiality of chemokines, including MCP-1, CCL15, CCL20, and CXCL14, as biomarkers for differential diagnosis between benign tumors and ovarian cancer (OC).
A cross-sectional study was conducted in women aged >18 years who had adnexal masses treated with elective surgery at the HRH Maha Chakri Sirindhorn Medical Center, Srinakharinwirot University, between 2020 and 2021. The preoperative MCP-1, CCL15, CCL20, and CXCL14 serum levels were measured using a sandwich enzyme-linked immunosorbent assay. Preoperative diagnosis was defined according to the risk of malignancy index. The histological diagnosis and cancer subtype were confirmed using pathological specimens.
Ninety-eight participants were preoperatively diagnosed with malignant tumors. The pathological diagnosis confirmed OC in 33 patients and disclosed 27 misdiagnosed cases, of which endometriotic cyst was the most common (44.44%). CCL20 and CA125 serum levels were significantly higher in the patients with cancer than in those with benign. In addition, CCL20 level could differentiate between benign and early-stage malignancy. Furthermore, only CCL20 levels could distinguish endometriotic cysts from OC, whereas CA125 levels could not. Concordant with the serum protein level, the increased mRNA level of CCL20 was observed in ovarian cancers comparing with that in benign tissues. We found that CCL20 levels could differentiate between benign tumors and OC with 60.61% sensitivity and 75.44% specificity at the optimal cutoff value of 38.79 pg/ml. Finally, the logistic regression model integrating CCL20, CA125, and menopause status promoted diagnostic accuracy by increasing the specificity to 91.23%.
Our study revealed the potential usefulness of CCL20 level as a biomarker for diagnosing early-stage OC with endometriosis differentiation. We recommend further studies to confirm the accuracy of CCL20 levels with the current diagnosis in a large patient sample.
Ovarian cancer (OC) is the second most common gynecological cancer, with a reported incidence rate of 7.9 per 100,000 women annually (1). It is the leading cause of cancer death among women in developed countries, including Thailand (2, 3). Its fatality and 5-year survival rates are approximately 70% and 28%, respectively (4). Almost 75% of OC cases are diagnosed in the advanced disease stage (III/IV), when transperitoneal, hematogenous, and/or lymphatic spread have already occurred, leading to poor prognosis and high recurrence rates (5). Epithelial OC (EOC) is the most common type that can be classified histologically as high- and low-grade serous, mucinous, clear cell, and endometrioid carcinomas (1). The standard treatment for OC is surgical removal of the tumor, followed by first-line platinum-based and paclitaxel chemotherapy (6). Although serum carbohydrate antigen 125 (CA125) and ultrasonography are the most widely used diagnostic tools for OC, they demonstrate low sensitivity and specificity for the early detection of OC. Elevated CA125 levels are not specific to OC and also occur in non-cancer patients with endometriosis inflammation and patients with other cancer types (7, 8). In addition, approximately 20% of patients with OC have normal CA125 levels (9). Therefore, the current research focuses on novel diagnostic biomarkers that show high accuracy and specificity for OC (10, 11). Inflammatory conditions increase the risk of developing OC (12). Inflammation can activate transcription factors such as nuclear factor-κB, signal transducer and activator of transcription 3, and hypoxia-inducible factor 1α, which are associated with inflammatory mediators, prostaglandins, cytokines, and chemokine production in tumor cells. These processes activate various inflammatory cells and change the tumor microenvironment, promoting malignant generation (13). Chemokines are small cytokines secreted during the inflammatory process to regulate immune cells. Chemokines, which consist of four conserved cysteine residual domains linked by a disulfide bond, are divided into four groups according to the position of these cysteines: C, CC, CXC, and CX3C (14, 15). Chemokines can induce the migration of leukocytes, including monocytes, lymphocytes, granulocytes, natural killer cells, and tumor-associated macrophages (TAMs), into pathological conditions, including infections, inflammatory reactions, and tumors. Chemokines affect tumor progression via several mechanisms, including angiogenesis, cell proliferation, migration, and immune invasion, which are necessary to ensure the success of growing tumors and disseminating metastasis (15). Angiogenesis is an important condition in the progression of cancers, including ovarian carcinoma (16). Several CXC chemokines have been reported to induce either endothelial cell migration or proliferation and neovascularization and to play roles in tumor growth. CC chemokines have been reported to play both direct and indirect roles in angiogenesis (17). The angiogenic response induced by CCL2 was accompanied by an inflammatory response and induced chemotaxis of human endothelial cells and the formation of blood vessels (18). Several chemokines have been explored for their potential usefulness in the diagnosis and prognosis of OC (19, 20). Previous studies have shown that elevated levels of some chemokines, including CCL2 or MCP-1 (21), CCL15, CCL20 (22), and CXCL14 (23), were detected in serum samples from patients with OC. However, these potential biomarkers have never been validated as preoperative diagnostic markers. The aim of this study was to evaluate the diagnostic accuracy and optimal cutoff values of CCL2, CCL15, CCL20, and CXCL14 compared with CA125 for OC diagnosis.
This cross-sectional study was conducted at the Department of Obstetrics and Gynecology, HRH Maha Chakri Sirindhorn Medical Center, Srinakharinwirot University, with the approval of the institutional review board (certificate No. SWUEC/F-047/2563). Written informed consent was obtained from all participants. All procedures were performed in accordance with the relevant guidelines and regulations. Research staff members had completed the Good Clinical Practice training.
The inclusion criteria were women with clinically diagnosed ovarian or adnexal masses who underwent elective surgery between July 2020 and October 2021, with ages >18 years, and with an American Society of Anesthesiologists physical status score of I–II. Participants were excluded if they had a history of chemotherapy or radiotherapy, serious psychiatric disease, pregnancy, previous history of OC or any malignancy, active inflammatory disorders, i.e., rheumatoid arthritis, multiple sclerosis, and type I diabetes, and refused to sign an informed consent form. Patients without intraoperative findings indicating a pelvic or adnexal mass were also excluded from the data analysis. All participants were hospitalized for preoperative preparation at least 24 hours before surgery, and a blood sample was preoperatively collected via a peripheral venous puncture. Clinical data, preoperative diagnosis, operative procedure, and postoperative diagnosis were recorded. Preoperative diagnosis was defined according to the risk of malignancy index (RMI) based on the score calculated from the serum CA125 level, menopause status, and ultrasonography. Postoperative diagnosis was based on pathologists’ histological interpretations.
The clot blood sample was centrifuged at 3,000 RPM for 10 minutes to separate serum immediately and stored at −20°C until the analysis. Chemokines were quantitatively analyzed in patient serum using a sandwich enzyme-linked immunosorbent assay (ELISA) kit from R&D Systems (catalog Nos. DY279, DY360, DY363, and DY866 for MCP-1, CCL20, CCL15, and CXCL14, respectively), in accordance with the manufacturer’s instructions. The absorbance was determined instantly using a CLARIOstar Plus microplate reader (BMG LABTECH, Germany) at wavelengths of 450 and 570 nm. The concentration of each chemokine was calculated by subtracting each absorbance reading at 570 nm from the reading at 450 nm and the average of duplicate readings, and then normalized with the average of the blanks. A standard curve was created with a quadratic polynomial fitting curve in Microsoft Excel for MAC version 16.58. The standard curve was considered as best fitted if r 2 value of the fitting line was >0.98. The CA125 level was obtained from routine hospital laboratory measurements.
The mRNA expression levels of CCL20 and CA125 (MUC16) in human ovarian tissues were retrieved from two public datasets, GSE4122 (24) and GSE17308 (25), available in NCBI’s Gene Expression Omnibus, which included 46 and 67 patients, respectively. We categorized the patient tissues, based on the pathological diagnosis, into 3 groups, i.e., benign, borderline ovarian tumors (BOTs), and malignancy groups.
Statistical analysis was performed using a power of 80% and an α value of 0.05. Chemokines with concentrations below the detection limit were excluded from the data analysis. The patients’ baseline characteristics were presented as frequency and percentage or mean and standard deviation. Categorical data, including patient and operation characteristics, were analyzed using the chi-square test. Receiver-operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) was calculated to compare the performance of the biomarkers for predicting OC in terms of sensitivity, specificity, positive predictive value, and negative predictive value. The ROC curves were compared using a chi-square test. The web tools easyROC version 1.3.1 and STATA 14.1 (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP) were used to calculate the optimal cutoff values (Youden index) (26). A logistic regression analysis was performed for the preoperative factors. The preoperative factors were integrated into the model if the model’s parameter p-value was <0.05 and excluded from the model if the p-value was >0.1. P-values <0.05 were considered statistically significant. A statistical analysis was performed using GraphPad Prism version 9.3.1 (350) for macOS Monterey (GraphPad Software, San Diego, California USA, https://www.graphpad.com) and Statistical Package for Social Sciences (SPSS) versions 28 for Windows (SPSS, Chicago, IL, USA).
This study included 100 patients for eligibility. Two patients with gastrointestinal stromal tumors were excluded from the analysis; thus, 98 patients were included in the analysis, as shown in Table 1 . We used the RMI to determine the preoperative risk of malignancy. Forty patients (40.82%) were preoperatively diagnosed with benign disease; and 58 patients (59.18%), with malignancy. However, the pathological diagnosis indicated that 59 patients (60.20%) had benign tumors, 6 (6.13%) had BOTs, and 33 (33.67%) had OCs. We found a discrepancy between the preoperative and postoperative diagnoses in 27 patients (27.55%), and 23 histologically confirmed benign cases were preoperatively diagnosed as malignant. Endometriotic cyst was the most common misdiagnosis, with up to 12 misdiagnosed patients (44.44%).
To determine the preoperative protein levels in the patients’ serum samples, an ELISA kit was used to measure the concentrations of MCP-1, CCL15, CCL20, and CXCL14 chemokines. A total of 98 samples were analyzed for MCP-1, CCL15, and CXCL14. Two samples had CCL20 concentrations below the detection limit, leaving 96 samples analyzed for this chemokine. The concentrations of MCP-1 and CCL20 in the preoperatively diagnosed malignant group were significantly higher than those in the benign group (median [IQR], pg/ml: MCP-1, 75.46 [23.772–92.71] vs 37.67 [0.65–85.58], p=0.041 and CCL20, 36.83 [22.58–65.76] vs 24.35 [8.84–40.19], p=0.009; Figures 1A, B ). CCL15 and CXCL14 levels were not statistically different between the malignant and benign groups ( Figures 1C, D ). As expected, the CA125 level (U/ml) significantly increased in the malignant group compared with the benign group (97.70 [5.65–245.80] vs 41.30 [15.88–87.43], p=0.0002; Figure 1E ). Next, we analyzed the chemokine concentrations in the samples categorized according to histologically confirmed diagnosis to identify the potential biomarker that could predict pathological results before surgery ( Table 2 ). The results showed that among the 4 chemokines, only CCL20 showed significantly different levels (pg/ml) between the benign and malignant groups (25.64 [13.59–39.32] vs 47.47 [23.69–93.30], p=0.015; Figure 1F ). As a control, the CA125 level (U/ml) was significantly different between the benign and malignant groups (51.70 [23.70–117.00] vs 122.00 [58.80–329.50], p=0.021; Figure 1G ). However, the chemokines and CA125 could not differentiate the BOTs in this study ( Table 2 and Figures 1F–J ).
To disclose the potential biomarkers that can distinguish between patients with malignancy and endometriotic cysts, chemokine levels were further analyzed in malignant cases compared with endometriotic and other benign cases, which were classified according to pathological results ( Figure 2 ). By comparing with other benign cases (median [IQR], U/ml: 29.00 [17.93–53.15]), CA125 levels (U/ml) were found to be significantly elevated in the endometriotic cysts and malignant cases (82.70 [51.00–163.00], p=0.015 and 122.00 [58.80–329.50], p=0.001, respectively) and were not statistically different between the endometriotic cysts and malignant cases ( Figure 2A ). Of interest, the CCL20 level (pg/ml) in endometriosis (24.86 [9.74–35.00]) was comparable with that in other benign cases (32.12 [14.33-40.20]) and significantly decreased compared with that in malignant cases (47.47 [23.69–93.30]), with a p-value of 0.019 ( Figure 2B ). In consideration of the misdiagnosed subgroup, the endometriotic cysts exhibited remarkably increased CA125 levels (U/ml: 183.50 [48.90–394.30]) compared with the other benign cases (46.00 [19.00–92.00], p=0.002), but the CCL20 levels (pg/ml) were not significantly different (32.28 [22.05–38.45] vs 34.50 [1.15–65.44]), respectively; Figure S1 ). These results suggest that high CA125 levels in endometriotic cases can lead to misdiagnosis, and measurement of CCL20 levels can potentially solve this problem.
To investigate whether CCL20 could differentiate between benign and early-stage malignancy, the CCL20 levels were subgroup analyzed by tumor stage, i.e., early stage (stages I and II) and advanced stage (stages III and IV). CCL20 levels (pg/ml) significantly increased in early-stage tumor than in benign cases (median [IQR], 53.02 [22.97–93.43] and 25.64 [13.59–39.32], p=0.045, respectively) ( Figure 3 ). CA125 level demonstrated no statistically significant difference between benign cases and early-stage OCs (median [IQR], 51.70 [23.70–117.00], and 132.4 [4.79–302.40], respectively) ( Figure S2 ).
We further validated our finding using two public patient datasets which had available data of the mRNA expression level of CCL20 and pathological diagnosis results. GSE4122 dataset showed that CCL20 level significantly increased in the malignancy compared to benign groups (p<0.001; Figure 4A ), in contrast to CA125 level which significantly decreased in the malignancy compared to the benign groups (p<0.0001; Figure S3A ). GSE17308 dataset showed a trend of high CCL20 level in malignancy, though the difference between benign and malignancy did not achieve the significance level ( Figure 4B ), whereas CA125 level increased in malignancy compared to benign groups ( Figure S3B ).
A ROC analysis was performed to demonstrate the pathological diagnostic performances of CA125 and CCL20 levels as potential biomarkers of OC ( Figure 5 ). The areas under the ROC curve of CA125 and CCL20 levels for differentiating malignant from benign (n=90; excluded 6 BOTs and 2 samples below the detection limit) were 0.666 (95% confidence interval [CI], 0.544–0.788) and 0.677 (0.558–0.796), respectively ( Figure 5A ). The areas under the 2 ROC curves were not significantly different (p=0.895). The optimal cutoff value defined using the Youden index method, to distinguish between benign and malignant tumors was 62.65 U/ml for CA125 and 38.79 pg/ml for CCL20. The sensitivity, specificity, and positive and negative predictive values at these cutoff points of the ROC curves are shown in Table 3 . Concisely, CCL20 exhibited higher specificity (75.44% vs 56.14%), improved diagnostic accuracy (70.00% vs 63.33%), and lower sensitivity (60.61% vs 75.76%) compared with CA125 at the optimal cutoff point. Logistic regression was used to build a diagnostic model to identify preoperative factors that could improve diagnostic accuracy. The model proposed three predictors, namely postmenopausal status, CA125 level, and CCL20 level, with odds ratios (95% CI) of 5.85 (1.99–17.18), 3.63 (1.22–10.79), and 3.12 (1.11–8.77), respectively. The model demonstrated 51.52% sensitivity and 91.23% specificity, with 76.76% diagnostic accuracy Table 3 and Figure 5B ).
The standard diagnosis of OC is based on serum CA125 levels, ultrasonography, and menopause status. Three of the 4 patients were diagnosed in the advanced stage and had a poor prognosis and increased mortality rate compared with early-stage patients (4–6). The CA125 level is the most widely used biomarker for diagnosis, monitoring treatment efficacy, and predicting the prognosis of OC (27). However, the CA125 level is not reliable for screening or early detection because of the high rates of false-positive and false-negative results and the many factors that affect CA125 levels, such as age, race, ethnicity, smoking history, and obesity (28). Elevated CA125 levels could be found in patients with benign conditions (29); this raises the malignancy risk (RMI score) and leads to misdiagnosis. Of the 27 misdiagnosed patients in this study, 12 who had endometriosis (44.44%) were preoperatively diagnosed with malignancy and exhibited significantly high serum CA125 levels compared with those with other benign cases ( Figure S1A ). The percentage of misdiagnosed endometriosis in our study was comparable with that previously reported by Yamamoto et al. (30), who found endometriosis in 40% of false-positive cases using the RMI method. This non-specificity drawback of CA125 warrants the need for novel OC-specific diagnostic biomarkers. CCL20 is a proinflammatory chemokine originating from T-helper 17 cells that is responsible for normal function of lymphocyte cells. CCR6 is a CCL20 receptor expressed on dendritic cells, T cells, and B cells (31). In cancer research, CCL20 plays a crucial role in neoplastic processes, and TAM is its major source. CCL20 and CCR6 in tumors can promote cell proliferation, migration, invasion, metastasis, and angiogenesis by directly stimulating vascular endothelial cells, thereby increasing vascular endothelial growth factor expression (32–34). CCL20 levels are elevated in many cancer types such as breast, liver, and pancreatic cancers, but are low in adrenal gland and lung cancers (31). In OC, it has been reported to contribute to promoting chemotherapy resistance, stimulating migration, and poor disease prognosis (35–37). Increased CCL20 levels were found in serum samples from patients with recurrent OC who were completely treated with platinum-based chemotherapy and correlated with decreased recurrence-free survival rates (19). Our results showed that CCL20 levels could distinguish OC from benign diseases, including endometriotic cysts. The misdiagnosed patients with endometriosis demonstrated low serum CCL20 levels, similar to those in other benign cases, but showed remarkably high CA125 levels ( Figure S1B ). This results in improved specificity (75.44%) and diagnostic accuracy (70.00%) of CCL20 compared with those of CA125 ( Table 3 ). Serum CA125 levels are elevated in women with cystic ovarian endometriosis and imitate those in OC, especially in premenopausal women (38, 39). By contrast, the serum CCL20 levels in premenopausal women have been reported to be lower in those with endometriosis than in those without endometriosis (40). Our results indicate the potential of usefulness of CCL20 level as a biomarker to differentiate endometriosis in the setting of ovarian tumors with elevated CA125 levels, which can reduce the false positivity rate from 41.11% in preoperative diagnosis to 15.56% at the optimal cutoff value of CCL20. This suggests that the factors used for preoperative diagnosis should be modified to improve diagnostic accuracy. Moreover, the CCL20 levels significantly increased in early-stage malignancy compared to benign. Whereas CA125 has low specificity to detect the early-stage OC in our study and in previous publications (41–43). These suggest the advantage of using CCL20 level as a biomarker for early OC detection. We further validated the plasma protein level of CCL20 using differential gene expression analysis in benign and malignant tissues. Although there was a limited number of patient tissues, the CCL20 gene expression was significantly high in malignant tissues comparing with that in benign tissues. These consistent expressions support the theory that CCL20 is a proinflammatory cytokine which is excreted by tumor cells and involved with early tumor progression (44). The accuracy of the diagnostic biomarker was calculated using ROC analysis, and the results suggested that the optimal cutoff value of CA125 was 62.65 U/ml, with 75.76% sensitivity and 56.14% specificity ( Table 3 ). Maggino et al. (45) reported that the cutoff value of CA125 was 65 U/ml, with 71.7% sensitivity and 92.5% specificity. CA125 showed lower sensitivity (60%) and specificity rates (89%) at a cutoff value of 65 U/ml in a subgroup of premenopausal patients (46) compared with postmenopausal women (78% sensitivity and 97% specificity). A possible explanation for the low specificity of CA125 in our study is that almost 70% of our population were premenopausal and up to 60% had benign disease. The ROC analysis revealed that using either CA125 or CCL20 could not differentiate OC in all cases. Thus, we generated a logistic regression model that finally integrated three significant predictors of OC, namely postmenopausal status and CA125 and CCL20 levels at optimal cutoff values. The model combining these three parameters exhibited high specificity (91.23%), with a sensitivity of 51.52% and an accuracy of 76.76%. Postmenopausal status had the highest probability (5.85 times) of predicting OC, followed by CA125 and CCL20 levels (3.63 and 3.12 times, respectively). The final model increased the probability of diagnosing OC to up to 11 times, indicating that the combination of CCL20 measurements and the diagnostic index led to a more efficient diagnosis with improved specificity to OC than the standard RMI. These results indicate the diagnostic potential of CCL20 as a specific biomarker and preoperative diagnostic tool for OC. To our knowledge, CCL20 has never been studied for its role as a diagnostic biomarker for OC. Owing to the limited number of patients with OC in this study (n=33), we did not find any candidate biomarkers for the histological subtypes of OC (data not shown). The only tissue subtype that could be differentiated was the tissue origin, that is, epithelial, or nonepithelial cells, based on the elevated CA125 levels in the epithelial subtype ( Figure S4 ). This indicates the variability of CA125 levels among OCs, which could be further studied. Although only CCL20 showed potential as an upregulated biomarker for OC, all other chemokines tested in this study demonstrated trends of high serum levels in malignancy. We recommend performing further research on CCL20 with a larger sample size to confirm our findings and explore the new aspect of this cytokine, especially in women with endometriotic cysts. To investigate the additional role of CCL20 as a biomarker for predicting prognosis and treatment efficacy, we suggest further study to monitor the postoperative levels of chemokines, disease progression of patients, and treatment efficacy. Lastly, BOTs are still a problem in differential diagnosis; the chemokines or CA125 could not differentiate them from benign or malignant diseases, which suggests the need for an efficient diagnostic method to overcome this difficulty. In conclusion, CCL20 and CA125 could be utilized as diagnostic biomarkers for OC, although CCL20 provides higher specificity to endometriotic disease, and early stage of OCs detection. However, the expression levels of CXCL14, CCL15, and MCP-1 are not suitable for predicting endometriosis, as they showed no significant difference between benign and malignant ovarian tumors. This finding must be validated in a larger number of patients with histologically confirmed OC. Testing for a novel biomarker in patients before surgery will be beneficial for choosing the most appropriate therapeutic options.
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .
The studies involving human participants were reviewed and approved by Institutional Review Board (IRB), Strategic Wisdom and Research Institute, Srinakharinwirot University. The patients/participants provided their written informed consent to participate in this study.
WS, WW, PV, and CC performed formal analysis and investigation. WS also performed data curation, statistical analysis, and visualization. WS wrote the first draft of the manuscript. VY supervised formal analysis and investigation. PN contributed the resources. VY and PN contributed to the conceptualization, methodology, design of the study, validation of the results, funding acquisition, and reviewing and editing of the manuscript. All authors contributed to the article and approved the submitted version.
This study was granted by Research Fund, the Faculty of Medicine, Srinakharinwirot University and supported by The Second Century Fund (C2F), Chulalongkorn University
All members of Department of Obstetrics and Gynecology, Faculty of Medicine, Srinakharinwirot University. WS and VY are supported by The Second Century Fund (C2F), Chulalongkorn University.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647057 | Mi Lin,Ning-Zi Lian,Long-Long Cao,Chang-Ming Huang,Chao-Hui Zheng,Ping Li,Jian-Wei Xie,Jia-Bin Wang,Jun Lu,Qi-Yue Chen,Ya-Han Li,Zhu-Huai Peng,Xiao-Yu Zhang,Yi-Xian Mei,Jian-Xian Lin | Down-regulated expression of CDK5RAP3 and UFM1 suggests a poor prognosis in gastric cancer patients | 27-10-2022 | gastric adenocarcinoma,CDK5RAP3,UFM1,AKT pathway,prognosis | Purpose The relationship between the CDK5RAP3 and UFM1 expression and the prolonged outcomes of patients who underwent gastric cancer (GC) surgery was investigated. Methods Single-sample gene set enrichment analysis (ssGSEA), unsupervised clustering and other methods were used to verify the relationship between CDK5RAP3 and UFM1 in GC through public databases. Additionally, CDK5RAP3 and UFM1 expression in cancerous and paracancerous tissues of GC was analysed in the context of patient prognosis. Results CDK5RAP3 and UFM1 expression was downregulated synchronously, the interaction was observed between the two proteins, and UFM1 and CDK5RAP3 expression was found to be inversely associated to AKT pathway activation. Prognostic analysis showed that the prognosis is poorer for low CDK5RAP3 and UFM1 patients, than for high CDK5RAP3 and/or UFM1 (p<0.001) patients, and this expression pattern was an independent predictor for overall survival of GC. Coexpression of CDK5RAP3 and UFM1 combined with TNM staging can improve the accuracy of prognosis prediction for patients (p <0.001). Conclusions It is confirmed in our findings that a combination of CDK5RAP3 and UFM1 can produce a more precise prediction model for GC patients’ survival. | Down-regulated expression of CDK5RAP3 and UFM1 suggests a poor prognosis in gastric cancer patients
The relationship between the CDK5RAP3 and UFM1 expression and the prolonged outcomes of patients who underwent gastric cancer (GC) surgery was investigated.
Single-sample gene set enrichment analysis (ssGSEA), unsupervised clustering and other methods were used to verify the relationship between CDK5RAP3 and UFM1 in GC through public databases. Additionally, CDK5RAP3 and UFM1 expression in cancerous and paracancerous tissues of GC was analysed in the context of patient prognosis.
CDK5RAP3 and UFM1 expression was downregulated synchronously, the interaction was observed between the two proteins, and UFM1 and CDK5RAP3 expression was found to be inversely associated to AKT pathway activation. Prognostic analysis showed that the prognosis is poorer for low CDK5RAP3 and UFM1 patients, than for high CDK5RAP3 and/or UFM1 (p<0.001) patients, and this expression pattern was an independent predictor for overall survival of GC. Coexpression of CDK5RAP3 and UFM1 combined with TNM staging can improve the accuracy of prognosis prediction for patients (p <0.001).
It is confirmed in our findings that a combination of CDK5RAP3 and UFM1 can produce a more precise prediction model for GC patients’ survival.
As a well-known malignant tumor, gastric cancer leads to high lethality of patients worldwide, which makes it rolling as a leading cause to death. The latest epidemiological survey showed that gastric cancer ranks fifth and third global incidence and mortality rates, respectively, among malignant tumours. Globally, more than one million new gastric cancer cases are diagnosed every year, and approximately 800,000 people die of gastric cancer (1). At diagnosis and therapy, majority of patients are already at an advanced stage because of low specificity of early gastric cancer symptoms. Advanced gastric cancer patients have an unfavourable prognosis, with just around a 15% 5-year survival rate (2).. Accurate prognostic assessment helps to formulate reasonable treatment plans and follow-up plans. The TNM staging system is the foremost predictor for gastric cancer prognosis. However, even with the same TNM stage, the prognosis of patients is not the same. In 2014, data from the Gastric Cancer Genome Atlas Research Network confirmed the molecular heterogeneity of gastric cancer (3). Therefore, the prognostic evaluation of the biological potential of gastric tumours has attracted increased attention. It has vital theoretical and clinical significance for the prognostic evaluation of gastric cancer to explore the molecular markers for early identification of gastric cancer and the important role of molecular targeted therapy. CDK5RAP3, known as C53, is an activation binding protein of cyclin-dependent kinase 5; it contains 506 amino acid residues and has a zinc-leucine zinc finger structure (4). CDK5RAP3 plays a key role in the formation and evolution of various malignant tumours (5). In our previous researches, we found that CDK5RAP3 can inhibit the phosphorylation of AKT in gastric cancer (6), thereby inhibiting the GSK-3β mediated phosphorylation, degrading β-catenin and acting as a tumour suppressor in the occurrence and progression of gastric cancer (7). UFM1, a amall ubiquitin protein that contains 85 amino acids, was first discovered by Komatsu in 2004. UFM1 is first activated by UBA5 and is then converted into UFC1 and UFL1. UFL1 recognizes and helps UFM1 to bind the target protein. Finally, UFM1 processes and modifies the target protein to perform its biological vital activities. UFM1 and its modification system participate in different pathophysiological and biological processes, including the cell cycle, fatty acid β oxidation, cell survival, and hypoxia tolerance (8–10). Research has demonstrated that the development of breast cancer involves UFM1 (11). In previous studies, we found that UFM1 can also negatively regulate PI3K/AKT signalling by increasing the ubiquitination of PDK1 to inhibit the invasion and metastasis of gastric cancer (12). The Akt-related signal transduction pathway is a complex signalling network mediated by growth factor receptors (GFRs) (13). Activation of this pathway suppresses cell apoptosis triggered by different stimuli, increases progression and proliferation of the cell cycle, participates in the neovascularization, plays an important role in the formation of tumours, and participates in invasion and metastasis of tumours (14–16). Thereby, we considered that CDK5RAP3 and UFM1 may play a coordinated role in inhibiting the gastric cancer invasion and metastasis. Although some studies have suggested that UFM1 binds to CDK5RAP3, the expression of the two proteins and their effects on the prolonged survival in gastric cancer have not been documented yet. Therefore, we investigated the correlation between UFM1 and CDK5RAP3 expression and the prognosis of gastric cancer using public databases. We also detected the expression of the two indicators in 215 gastric cancer tissue samples using IHC, Western blotting and qPCR. To improve the accuracy of judging prognosis in gastric cancer, the relationship between expression of these two proteins and relevant clinical and pathological characteristics, as well as long-term survival in patients was analysed.
We searched the published gastric cancer gene expression database systematically, including those with complete clinical information and excluding those with no survival information. Finally, we gathered The Cancer Genome Atlas Stomach Adenocarcinoma (TCGA-STAD) cohorts and 17 Gene Expression Omnibus (GEO) cohorts of samples from patients with GC for this study (GSE54129, GSE65801, GSE35809, GSE51105, GSE13861, GSE27342, GSE29272, GSE63089, GSE19826, GSE79973, GSE13911, GSE51575, GSE118916, GSE122401, GSE130823, GSE15459, GSE66229) and the TCPA database for analysis. The original data were collected and downloaded from GEO (http://www.ncbi.nlm.nih.gov/geo/), TCGA (https://portal.gdc.cancer.gov/), and TCPA (https://www.tcpaportal.org/tcpa).
The tissues in this study were selected from gastric adenocarcinoma tissue specimens of 215 patients undergoing radical gastrectomy for gastric cancer in our center from January 2013 to December 2014. All patients were newly diagnosed and before surgery they had not received chemotherapy or radiation treatment. The patients were pathologically confirmed to have gastric adenocarcinoma after surgery with comprehensive clinicopathological information. The data were analyzed retrospectively. This study was approved by the Fujian Medical University Union Hospital Ethics Committee and written permission was obtained from every relevant patient.
We obtained 3 GFR gene sets (KRAS_SIGNALING_UP and AKT_UP. V1_DN and MTOR_UP. V1_DN) from C6 (oncogenic gene sets) of MSigDB (https://www.gsea-msigdb.org/). Using the R software package “GSVA” (gene set variation analysis for microarray and RNA-seq data), we scored each sample in the TCGA cohort by ssGSEA (method = “ssgsea”, ssgsea.norm = TRUE, verbose = TRUE).
Unsupervised clustering methods (K-means) were used to classify the TCGA cohort into different clusters based on the enrichment of GFR pathways. The clustering factors were the ssGSEA scores of the three GFR gene sets. These scores were first converted to z scores to improve the accuracy of clustering. We determined the final number of clusters according to the algorithm provided by the R software package “NbClust”. Finally, the TCGA queue was accurately divided into 3 clusters defined as Cluster A, Cluster B, and Cluster C.
We performed GSEA on the TCGA and GEO datasets (GSE54129, GSE65801, GSE35809, and GSE51105). First, we used the mean ± standard deviation (SD) of the CDK5RAP3 expression value as the cut-off point to divide each data set into three groups: the group of high, moderates and low. Next, we compared the high and low expression group to obtain differentially expressed genes. Additionally, the R package “clusterProfiler” (v3.12.0)0 (https://guangchuangyu.github.io/software/clusterProfiler) was applied to perform GSEA on these differential genes. MSigDB provided us with all of the hallmark and oncogenic gene sets (https://www.gsea-msigdb.org/).
Tumour specimens containing enough formalin-fixed and embedded by paraffin were sliced into 4-μm serial sections and mounted for immunohistochemical analysis on silane-coated glass slides. The sections were dewaxed, rehydrated, antigen repaired, blocked and then incubated with appropriate antibodies. The rabbit anti-human CDK5RAP3 (ab24189; 1:200; Abcam) or UFM1 (ab109305; 1:200; Abcam) antibody was used as the primary antibody.
Two experienced pathologists independently assessed IHC-stained tissue slices and scored them based on the intensity of cell staining and the positive ratio of the stained tumour cells. The proportion and intensity of CDK5RAP3-positive and UFM1-positive cells in random selection visual areas were evaluated to indicate the protein expression level. The following were the staining score standards for CDK5RAP3 and UFM1: no staining was indicated by a score of 0; the light yellow was defined as mild staining with a score of 1; the yellowish brown was defined as moderate staining with a score of 2; the brown was defined as significant staining with a score of 3. The following were the proportional score standards for stained tumor cells: when less than or equal to 5 percent cells were positive, the score was 0; when the positive cells were range from 6 to 25 percent, the score was 1; when the positive cells were range from 26 to 50 percent, the score was 2; when the positive cells were greater than or equal to 50 percent, the score was 3. ( Figure S1 ). The final score ranging from 0 to 9 for the expression of CDK5RAP3 and UFM1, was obtained by multiplying the staining score and proportional score. The low-expression group was defined as patients having a final score <4. The high-expression group included the remaining patients.
We cut fresh soy-sized gastric cancer tissue and paracancerous tissue pieces into a shaking tube. Next, lysis solution was added (1 mg of tissue plus 6 µl of lysis solution). The lysis solution comprising RIPA lysis solution (Thermo Fisher Scientific, Waltham, MA, USA) + PMSF solvent + Cocktail (Roche, South San Francisco, CA, USA) was prepared (100:1:1). The tubes were then placed in the oscillator at 5 m/s for 30 s. Thereafter, the samples were subjected to shaking after 12000 rpm 4 times, followed by centrifugation for 5 min. The supernatant was then pipetted into a new EP tube. The protein concentration was measured by the BCA method, and the protein sample (loading volume per well 40 μg) was separated by 10% SDS-PAGE and transferred to a PVDF membrane. Subsequently, the membrane was blocked with 5% skim milk for 1 hour at room temperature. Next, the membrane was incubated with primary antibodies (CDK5RAP3, UFM1 and GAPDH) at 4°C overnight. After that, the membrane was washed with washing buffer (TBS-T) 3 times, 5 min each time, and then incubated with HRP secondary antibody (Cell Signaling Technology) for 1 h at room temperature. GAPDH was used as an internal control. Finally, the membrane was washed with TBS-T for 30 min and the protein bands were detected by enhanced chemiluminescence (Amersham Corporation, Arlington Heights, IL, USA). The following antibodies were used by Western blots: CDK5RAP3 (ab24189; 1:1000 dilution; Abcam, Cambridge, MA, USA), UFM1 (ab109305; 1:1000 dilution; Abcam, Cambridge, MA, USA), p-AKT (serine 473) (ab81283, 1:1000 dilution; Abcam, Cambridge, MA, USA) and GAPDH (#5174; 1:2000 dilution; Cell Signaling Technology).
Total RNA from gastric cancer and paracancerous tissues was extracted using Invitrogen’s TRIzol kit according to the manufacturer’s instructions and used to obtain cDNA using Takara’s reverse transcription system. The copy numbers of GAPDH, CDK5RAP3 and UFM1 were detected using qPCR. The following were the detailed primer sequences: CDK5RAP3 Forward primer: 5′-GCTGGTGGACAGAAGGCACT-3′ Reverse primer: 5′-TGTCCTGGATGGCAGCATTGA-3′ UFM1 Forward primer: 5′-GTCCCC AGCACACTAGAGGA-3′ Reverse primer: 5′-GGA AAAGAGCGGGAG AGAGT-3′ GAPDH Forward primer: 5′-GAAGGTGAAGGTCGG AGT-3′, Reverse primer: 5′-GAAGATGGTGATGGGATTTC-3′ GAPDH was used as an internal reference, and the ΔΔCt method was used for analysis.
Protein was extracted from stably transfected cells (HGC-27) overexpressing UFM1, and the BCA method was used to determine the protein concentration. A small amount of protein solution was saved and boiled with 2× SDS sample buffer and then frozen at -20°C for Western blot analysis. Next, an appropriate amount of UFM1 antibody was added to the remaining protein solution at a ratio of 100 µg of protein/1 µg antibody and incubated at 4°C with gentle shaking overnight. Protein A/G agarose beads (20 µl) were incubated at 4°C for 2–4 h and centrifuged at 4°C at 3000×g for 3 min. It discarded the supernatant and washed the agarose beads on 5 times with a buffer of 1 ml lysis. After the final removal of the supernatant, 20 µl of 2× SDS was added to the pellet, followed by boiling in water for 5 min. Finally, the CDK5RAP3 antibody was used for Western blot.
According to the institutional follow-up protocol, qualified doctors monitored all patients by outpatient clinics, phone calls, emails, letters or visits. The first 2 years of follow-up were completed every 3 months. The next 3 years of follow-up were completed every 6 months. Then they were followed up annually until death or after 5 years. Most of the patients had undergone physical exams, laboratory tests, imageological examinations and annual gastroscopy. The time from operation to last follow-up or death was defined as the overall survival time. The follow-up rate of the whole group was 93.56%, and the median follow-up time was 57 months (range, 2–83 months).
All statistical analyses were performed using the Social Science Statistical Software Package (SPSS) version 23.0 for Windows (IBM, Chicago, IL, USA) or R software (version 3.6.2). If not specified, the results were shown as percentages or means ± SD. As needed, the data were analysed by chi-square test, Fisher’s exact test or Student’s t test. The survival rate was evaluated by Kaplan-Meier method and log-rank test. The Cox proportional hazards model was used for univariate and multivariate prognostic analysis. Multivariate analysis was performed on factors with p<0.05 in univariate analysis. Statistical significance was indicated when the P value was less than 0.05. Pearson’s correlation or Spearman’s correlation was used to estimate the correlation coefficient (p <0.05). Additionally, the protein interaction network was constructed using GeneMANIA (http://www.genemania.org/). A receiver operating characteristic (ROC) curve and the area under the curve (AUC) were computed to assess discriminative ability.
First, we used unsupervised clustering methods to classify 375 tumour samples from The Cancer Genome Atlas (TCGA) database into three molecular subgroups (Cluster A, Cluster B, and Cluster C) based on the three characteristic pathways of GFRs: KRAS, AKT, and MTOR. The heat map showed that the downstream signalling pathway-related genes GFR signature, GF and GFR were inhibited in patients in Cluster A, while they were activated in patients in Cluster C ( Figure 1A ). By analysing the related proteins of the GFR pathway from The Cancer Proteome Atlas (TCPA) database, we observed that the GFR pathway-related proteins SYK, PDK1, P90RSK, 4EBP1, and BIM were found to be highly expressed in Cluster A, PREX was found to be highly expressed in Cluster B, and CKIT, AMPKALPHA, PKCALPHA_pS657, BAD_pS112, PKCALPHA, PACDELTA_pS664, SHP2542, TUBERIN_pT1462, and IRS1 were found to be highly expressed in Cluster C, with significant differences ( Figure 1B ). Survival analysis also indicated that the overall survival of the patients from Cluster C was lower than that of the patients from Cluster A (p = 0.043) ( Figure 1C ). To explore which genes played a key regulatory role in the GFR pathway, we compared the genetic changes in patients in Clusters B vs. A, C vs. A, and C vs. B. The Venn diagram showed that Clusters B vs. A, C vs. A and C vs. B had 507 common downregulated genes ( Figure 1D and Table S1 ), and 1,536 common upregulated genes ( Figure 1E and Table S2 ). Analysing the co-downregulated genes and CDK5RAP3-interacting proteins in the string database, we found that the CDK5RAP3 and UFM1 genes were included in the 507 common downregulated genes, and the mRNA levels of CDK5RAP3 and UFM1 in patients of category C were lower than those in patients of categories A and B. The log fold-change of CDK5RAP3 was -0.741 in Cluster C vs. Cluster A and -0.567 in Cluster C vs. Cluster B. The log fold-change of UFM1 was -0.636 in Cluster C vs. Cluster A and -0.423 in Cluster C vs. Cluster B. ( Figures 1F, G ). Additionally, an interaction was observed between the two proteins ( Figure 1H ).
We further performed pathway enrichment analysis of patients with high and low CDK5RAP3 expression in the TCGA and GEO databases. The mountain map, heat map and GSEA enrichment analysis map all indicated that CDK5RAP3 expression negatively correlated with AKT pathway activation ( Figures 2A–C ), a finding that was consistent with previous research results (6). Additionally, the correlation analysis of four GEO databases (GSE13861, GSE27342, GSE29272, GSE63089) and the TCGA database revealed that the expression levels of UFM1 and CDK5RAP3 were significantly correlated ( Figures 2D–H ). Co-IP experiments confirmed that UFM1 had a direct binding effect with CDK5RAP3 ( Figure 2I ). Therefore, we knocked down and overexpressed UMF1 and CDK5RAP3 in the HGC cell line to verify that UFM1 and CDK5RAP3 negatively correlated with AKT pathway activation. The results showed that knocking down UFM1 caused a decrease in CDK5RAP3 expression and reduced the inhibition of AKT phosphorylation, while the overexpression of UFM1 caused an increase in CDK5RAP3 to enhance the inhibition of AKT phosphorylation ( Figure 2J ). However, the UFM1 didn’t change when CDK5RAPS was knocked down or overexpressed ( Figure 2K ).
Analysis of 7 GEO databases (GSE13861, GSE54129, GSE19826, GSE79973, GSE13911, GSE51575, GSE29272) showed that CDK5RAP3 expression was low in gastric cancer ( Figure 3A ). CDK5RAP3 expression levels in cancerous and paracancerous tissues from 15 cases in GSE118916, 80 cases in GSE122401, and 47 cases in GSE130823 were found to be low ( Figure S2 ), as was UFM1 expression in cancerous and paracancerous tissues from 15 patients in GSE118916 ( Figure 3B ). Furthermore, we used samples from the internal centre for verification. IHC staining of cancerous and paracancerous tissues from gastric cancer patient showed that CDK5RAP3 and UFM1 protein expression in cancerous samples were both lower than that in paracancerous ( Figure 3C ). IHC staining score was used to analyse CDK5RAP3 and UFM1 protein expression in paraffin-embedded gastric cancer samples from 124 patients. CDK5RAP3 was found to be lowly expressed in 102 patients (82.3%) and had high expressions in 22 patients (17.7%). The expression levels of UFM1 were found to be low in 93 patients (75. 5%) and high in 31 patients (25.0%). Spearman’s correlation analysis indicated that CDK5RAP3 and UFM1 expression was significantly correlated ( Figure 3D ). We also used Western blotting to detect CDK5RAP3 and UFM1 expression in the cancerous and paracancerous tissues of 43 gastric cancer patients ( Figure 3E ) and simultaneously detected the mRNA levels of CDK5RAP3 and UFM1 in the tumour tissues of 48 patients with gastric cancer. Pearson’s correlation analysis showed that the expression of the two mRNA levels was positively correlated ( Figure 3F ).
The overall survival was reduced dramatically in patients with low CDK5RAP3 expression compared with patients with high CDK5RAP3 in the 3 GEO databases (GSE13861, GSE15459 and GSE66229) and the TCGA database ( Figure S3 ). Similarly, the overall survival rate was significantly worse among patients with low UFM1 than in patients with high UFM1 ( Figure S4 ). In the GSE66229 database, the patients with low CDK5RAP3 expression had a significant lower disease-free survival rate than that of patients with high CDK5RAP3 expression, and the patients with low UFM1 expression also had a significant lower disease-free survival rate than those with high UFM1 expression ( Figure S5 ). Regarding the internal centre data, the 3-year overall survival rate was 66.9% with median 57 months follow-up for the entire group. According to survival analyses, the 3-year cumulative overall survival rate of high CDK5RAP3 expression patients was significantly higher than that of low CDK5RAP3 patients (81.8% vs. 62.7%, p < 0.05, Figure 4A ); those with low UFM1 expression exhibited a lower 3-year overall survival rate than patients with high UFM1 expression (58.1% and 90.3%, respectively; p 0.05; Figure 4B ). We further analysed the prognostic value of the combination of CDK5RAP3 expression and UFM1 expression by Kaplan–Meier analysis. In comparison to the other groups of patients, patients with low expression levels of CDK5RAP3 and UFM1 had a poorer 3-year cumulative survival rate—only 54.9%—which was substantially below CDK5RAP3 high and/or UFM1 high expression patients ( Figure 4C ). After combing the groups, we found that patients with low CDK5RAP3 and UFM1 expression had a significantly worse prognosis than those with high CDK5RAP3 and/or UFM1 expression (88.1%) (p < 0.001; Figure 4D ).
Analysis of factors associated with the expression of CDK5RAP3 and UFM1 in gastric cancer tissues showed that the CDK5RAP3 and UFM1 expression significantly correlated with BMI, lymph node metastasis, depth of invasion and pathological TNM stage ( Table 1 ). Combining the low CDK5RAP3 and UFM1 expression, analysis of related factors showed that the low expression level of the two was related to tumour size, depth of invasion, lymph node metastasis and TNM staging ( Table 2 ). BMI, tumour size and TNM staging were further included in the logistic regression model. The results of multivariate analysis suggested that TNM staging was an independent factor related to the low expression of CDK5RAP3 and UFM1 (I+II vs. III: 95% CI: 1.128–5.755, p = 0.023).
Cox regression analyses were used to clarify the prognostic value of CDK5RAP3 and UFM1 expression. Based on the univariate analysis, overall survival was related to BMI, tumour size, TNM staging and combined CDK5RAP3 and UFM1 expression ( Table 3 ). Multivariate analysis indicated that the coexpression level of CDK5RAP3 and UFM1, as well as TNM stage were both independent predictive variables for patient prognosis with gastric cancer ( Table 3 ).
We compared the accuracy of CDK5RAP3 or UFM1 expression, as well as combined CDK5RAP3 and UFM1 expression and TNM staging, in predicting gastric cancer survival using ROC curve analysis. The combination expression of CDK5RAP3 and UFM1 was more accurate in predicting patient survival than either CDK5RAP3 or UFM1 expression on its own (AUC was 0.638, 0.584, and 0.596; 95% CI was 0.532–0.740, 0.473–0.688, and 0.490–0.702; p = 0.021, 0.172, and 0.104 for CDK5RAP3 + UFM1, CDK5RAP3 and UFM1 respectively). Additionally, combined CDK5RAP3 and UFM1 expression had a prognostic value that was similar to TNM staging (AUC: 0.651, 95% CI: 0.601–0.786, p = 0.001; Figure 5A ). Furthermore, compared with CDK5RAP3 or UFM1 combined with or without TNM staging, the coexpression of CDK5RAP3 and UFM1 combined with TNM staging further improved the prognostic prediction accuracy of patients (p < 0.001, Figure 5B ). Thus, the combination of CDK5RAP3 and UFM1 expression had a higher prognostic ability for overall survival in GC patients.You may insert up to 5 heading levels into your manuscript as can be seen in “Styles” tab of this template. These formatting styles are meant as a guide, as long as the heading levels are clear, Frontiers style will be applied during typesetting.
Gastric cancer remains the third leading cause of death in China despite improvements in diagnosis and therapy in recent years (5, 17). To better guide diagnosis and therapy, identifying specific biomarkers linked to gastric cancer prognosis may help improve the accuracy of gastric cancerprognostic assessment (18–20). Based on our previous study and an examination of public databases, this study found that the expression of the UFM1 and CDK5RAP3 genes are downregulated synchronously in gastric cancer patients with poor prognosis and that an interaction occurs between the UFM1 and CDK5RAP3 proteins. Therefore, we chose to evaluate UFM1 as a prognostic factor with CDK5RAP3. GFRs and their abnormal signal transduction are important mechanisms of tumorigenesis and development, and they have become hot topics of research in recent years (10, 21). Many studies have shown that the abnormal function of growth factors and their receptors is an important cause of tumour occurrence and development. Such growth factor receptors have tyrosine kinase activity and can regulate the activity of downstream signalling pathways through phosphorylation (22, 23). The PI3K/Akt signalling pathway plays an important antiapoptotic role. Abnormalities in Akt-related signalling pathways are also associated with the occurrence of various tumours (13, 24). Therefore, we used public databases to search for proteins related to the GFR signalling pathway and CDK5RAP3 and attempted to identify biological prognostic indicators for gastric cancer. It was suggested that UFM1 was positively correlated with CDK5RAP3 and its low expression was associated with poorer prognosis of gastric cancer. Previous studies have shown that multiple proteins related to CDK5RAP3 and UFM1 and their modification systems (such as UFC1 and UFL1) are closely related (14, 15, 25). The correlation analysis of multiple public databases in this study also proved that the CDK5RAP3 and UFM1 expression were found to be substantially linked. To date, few studies have investigated the combined expression levels of CDK5RAP3 and UFM1 and its prognostic significance in gastric cancer. Therefore, in patients with gastric cancer, we assessed the relationship between relevant clinicopathological parameters and overall survival by detecting the CDK5RAP3 and UFM1 expression levels. In univariate analysis, low CDK5RAP3 expression was linked to a poor prognosis, and high UFM1 expression was linked to a better survival rate in gastric cancer patients, indicating that both CDK5RAP3 and UFM1 play a tumour suppressor role in gastric cancer. Further analysis of related factors showed that the CDK5RAP3 and UFM1 coexpression was strongly linked to the invasive depth, lymph node metastasis and TNM stage, indicating that the two proteins are closely related to tumour invasion and migration in gastric cancer. Additionally, we found that the functions of CDK5RAP3 and UFM1 in gastric cancer were positively correlated. Patients with low CDK5RAP3 and UFM1 expression had the worst prognosis; if either of the two proteins showed high expression, patient’s prognosis was dramatically better. We considered that because CDK5RAP3 and UFM1 both played a role as tumour suppressor proteins, when one of the two proteins was highly expressed, the tumour suppressor effect in gastric cancer results in no difference in survival. When both proteins were expressed at a low level, the inhibition of the tumour was relieved, resulting in the poorest prognosis of all groups. Further analysis showed that the accuracy of prognostic analysis using CDK5RAP3 and UFM1 expression was closer to the accuracy of TNM staging prognostic analysis and higher than that of using CDK5RAP3 or UFM1 expression alone. Therefore, combination of the CDK5RAP3 and UFM1 expression can improve the capacity to forecast the survival outcomes of patients with gastric cancer. The TNM staging system has been identified as a major prognostic factor for the gastric cancer. It’s also a valuable foundation for the formulation of gastric cancer treatment. However, differences in the prognosis of the same stage patients persist. In this study, we combined the coexpression of CDK5RAP3 and UFM1 with TNM staging for prognostic analysis. In comparison to the conventional TNM staging’s forecast accuracy, combining CDK5RAP3 and UFM1 expression with TNM greatly improved the accuracy of predicting gastric cancer patient survival. This finding indicated that the coexpression level of CDK5RAP3 and UFM1 could increase the accuracy of gastric cancer prognostic evaluation. Maybe it is possible to build a more precise model combining CDK5RAP3, UFM1 and TNM staging to predict 5-year survival of gastric cancer after surgery. It is worth exploring in the subsequent research. As a result, in clinical practice, the coexpression of CDK5RAP3 and UFM1 can be used in cooperation with TNM staging to effectively guide treatment and follow-up of patients with gastric cancer. This study mainly explored the impact of the coexpression level of UFM1 and CDK5RAP3 on the clinicopathological parameters of gastric cancer patients and its prognostic significance, providing a preliminary basis for further research. Further investigation of how UFM1 and CDK5RAP3 regulate AKT pathway and whether CDK5RAP3 and UFM1 are associated with metastasis would be highly significant. Therefore, the elucidation of related mechanisms warrant further study.
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi.nlm.nih.gov/, TCGA-STAD, GSE54129, GSE65801, GSE35809, GSE51105, GSE13861, GSE27342, GSE29272, GSE63089, GSE19826, GSE79973, GSE13911, GSE51575, GSE118916, GSE122401, GSE130823, GSE15459, GSE66229 https://www.tcpaportal.org/tcpa, TCPA https://www.gsea-msigdb.org/, GSEA.
The studies involving human participants were reviewed and approved by Fujian Medical University Union Hospital Ethics Committee. The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
ML, L-LC, and J-XL conceived of the study; ML, N-ZL, and L-LC conducted the experiment and performed the major analysis; ML and N-ZL prepare the manuscript; J-XL, C-MH, C-HZ, PL, J-WX, J-BW, JL, Q-YC, Y-HL, Z-HP, X-YZ, and Y-XM helped to collect the data and revise the manuscript critically for important intellectual content; All authors contributed to the article and approved the submitted version.
This study was funded by Construction Project of Fujian Province Minimally Invasive Medical Center (NO.[2021]662). Scientific and technological innovation joint capital projects of Fujian Province (No. 2018Y9008). Fujian Provincial Health Technology Project (2018-1-40). China Scholarship Council (202108350068).
We are thankful to Ju-Li Lin, Hua-Long Zheng, Guang-Tan Lin and Fujian Medical University Union Hospital for managing the gastric cancer patient database.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The handling editor DP declared a past collaboration with the author L-LC.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647058 | Liyu Wang,Ying Feng,Anquan Huang,Jianming Shi,Qinying Zhang,Fan Zhu,Bin Lv,Fen Guo,Tianming Zou,Luyao Zhang | Case report: Significant response to PD-L1 inhibitor after resistance to PD-1 inhibitor in an advanced alpha-fetoprotein-positive gastric cancer | 27-10-2022 | positive alpha-fetoprotein,gastric cancer,PD-1,immunotherapy,chemotherapy | Alpha-fetoprotein-positive gastric cancer (AFPGC) is a type of gastric cancer with a high degree of malignancy. The disease is more common in the elderly, with a high prevalence in males and generally atypical clinical manifestations. For advanced patients, the current treatment options are limited and, to date, few cases of advanced AFPGC have been treated successfully with conventional chemotherapy. With the development of molecular biology and immunology, tumor immunotherapy offers more therapeutic options to patients with advanced gastric cancer. This study describes a case of advanced gastric cancer in a young woman with a high blood alpha-fetoprotein (AFP) level (>54,000 ng/mL). The patient showed initial promising results when programmed cell death-1 (PD-1) inhibitor treatment was combined with chemotherapy after systemic chemotherapy failed. When the disease progressed again after 129 days, adjustment of the treatment regimen to Atezolizumab in combination with Irinotecan and Surufatinib capsules achieved partial remission (PR). There were no immune-related pneumonia, myocarditis, or other adverse effects observed. The patient currently has an overall survival of more than 14 months. This case demonstrated that switching from PD-1 inhibitor to programmed cell death-Ligand 1 (PD-L1) inhibitor therapy may overcome potential resistance. It providing a reference for immunotherapy of patients with AFP-positive advanced gastric cancer. | Case report: Significant response to PD-L1 inhibitor after resistance to PD-1 inhibitor in an advanced alpha-fetoprotein-positive gastric cancer
Alpha-fetoprotein-positive gastric cancer (AFPGC) is a type of gastric cancer with a high degree of malignancy. The disease is more common in the elderly, with a high prevalence in males and generally atypical clinical manifestations. For advanced patients, the current treatment options are limited and, to date, few cases of advanced AFPGC have been treated successfully with conventional chemotherapy. With the development of molecular biology and immunology, tumor immunotherapy offers more therapeutic options to patients with advanced gastric cancer. This study describes a case of advanced gastric cancer in a young woman with a high blood alpha-fetoprotein (AFP) level (>54,000 ng/mL). The patient showed initial promising results when programmed cell death-1 (PD-1) inhibitor treatment was combined with chemotherapy after systemic chemotherapy failed. When the disease progressed again after 129 days, adjustment of the treatment regimen to Atezolizumab in combination with Irinotecan and Surufatinib capsules achieved partial remission (PR). There were no immune-related pneumonia, myocarditis, or other adverse effects observed. The patient currently has an overall survival of more than 14 months. This case demonstrated that switching from PD-1 inhibitor to programmed cell death-Ligand 1 (PD-L1) inhibitor therapy may overcome potential resistance. It providing a reference for immunotherapy of patients with AFP-positive advanced gastric cancer.
Gastric cancer with elevated serum AFP levels confirmed histopathologically after excluding hepatitis, cirrhosis, hepatocellular carcinoma, germ-cell malignancy, and other illnesses that may cause AFP is referred to as AFPGC (1). AFPGC is considered to be one of the most aggressive tumor subtypes in gastric cancer. It has been reported that AFPGC accounts for 2.3%~7.1% of total gastric cancer in Asian countries and about 15% of total gastric cancer in western countries (2). It was reported that patient age, TNM stage and curable surgery were found to be associated with overall survival. The younger AFPGC patients are prone to have a more detrimental prognosis. A high level of AFP is an independent prognostic risk factor for gastric cancer because AFP is not only a product of the tumor but also plays a crucial role in proliferation, apoptosis, and angiogenesis of AFPGC cells (3). AFP has been reported to have a suppressive effect on lymphocyte transformation, to enhance tumor cell proliferation through the HGF and c-Met pathway (4), and to increase angiogenesis via Vascular Endothelial Growth Factor (VEGF) expression (5). According to the World Health Organization (WHO) (2019) classification of gastric cancer, serum AFP may be elevated in several types of gastric adenocarcinoma, such as hepatoid adenocarcinoma and gastric adenocarcinoma with enteroblastic differentiation. Previous research has found that serum AFP levels are an independent risk factor impacting patient survival (6), and there is currently no effective treatment for AFPGC. Here, we report the case of a 37-year-old woman suffering from gastric cancer with an extremely high expression of serum AFP level (>54,000 ng/mL). The patient showed initial promising results when PD-1 monoclonal antibody (mAb) treatment was combined with chemotherapy after systemic chemotherapy failed. When the disease progressed again after 129 days, adjustment of the treatment regimen achieved PR.
In January 2021, a 37-year-old woman complained of right upper abdomen discomfort and mild tenderness in a prone posture. She also had felt a palpable mass in the right upper abdomen. Physical examination revealed an enlarged liver with a bulge and palpable mass in the right abdomen. The lower margin of the enlarged liver was at the right midclavicular line about the level of the navel. The patient had tenderness in the liver area, and the Numeric Rating Scales (NRS) score was 2 points. The patient had no tumor-related family history and denied having a chronic liver illness (hepatitis, cirrhosis, and primary liver cancer) or combined reproductive system tumors. After hospitalization, an enhanced Computed Tomography (CT) examination was performed to further evaluate the patient’s condition, which showed there were multiple nodular lesions in the liver with local gastric wall thickening ( Figure 1A ). Among the tumor markers, AFP levels were significantly elevated to more than 54,000 ng/ml (normal level: < 12 ng/ml), and the serum carcinoembryonic antigen (CEA) levels were 22.82 ng/ml (normal range: < 5 ng/ml) ( Figures 2A, B ). Gastroscopy revealed a giant crateriform ulcer within the stomach body ( Figure 3A ), and pathological examination showed a poorly differentiated adenocarcinoma. Immunohistochemical labeling revealed the presence of proficient mismatch repair (pMMR)/microsatellite stability (MSS) as well as deficiencies in Human Epidermal growth factor Receptor 2 (HER2) and EBV-encoded RNA (EBER) hybridization ( Figures 3B–D ). The patient was diagnosed with poorly differentiated gastric adenocarcinoma with liver metastases (HER2 negative). Meanwhile, the patient was a gastric cancer with elevated serum AFP levels. Hepatitis, liver cirrhosis and other diseases that may lead to elevated AFP were excluded. Therefore, this patient was an advanced AFPGC patient. The patient received first-line chemotherapy in a two-drug combination regimen of paclitaxel liposome for injection and S-1 (tegafur/gimeracil/oteracil potassium). After two sessions of therapy, the disease status was stable, as determined by imaging. Then, in the third course, the S -1 dose was increased to 60 mg orally twice daily from day 1 to day 14 and continued to be combined with 210 mg paclitaxel liposome for injection. However, following the third session of chemotherapy, the patient reported that the upper abdominal distension was worse than before, and her AFP levels was consistently greater than 54,000 ng/ml. After communicating with the patient and completing the informed consent form for immunotherapy, the patient began to receive 200mg Tislelizumab injections in combination with 200mg Oxaliplatin treatment on April 17, 2021, every 21 days as a course. After two courses of treatment, the patient’s abdominal distension was relieved. An abdominal CT scan indicated that the liver lesions were smaller than before, and gastric wall thickening was reduced. The imaging evaluation showed that the disease status was stable. The original regimen was continued for two courses, and imaging was performed after the fourth course to assess PR ( Figure 1B ). Regular reexamination showed a progressive decrease in AFP level to 7169 ng/ml and the CEA level dropped to within the normal range ( Figures 2A, B ). The patient’s symptoms of pain in the liver area had been relieved and subsequently the analgesic drugs were stopped. At 6 months post-diagnosis, the serum level of AFP had increased again to 21,520 ng/ml ( Figure 2A ). At the same time, CT revealed that the liver lesions had increased in size again ( Figure 1C ). The curative effect was evaluated as progressive disease (PD). Then, the patient was given third-line therapy of Tislelizumab combined with Apatinib 250 mg once a day for 3 cycles from August 23, 2021 to October 5, 2021. Unfortunately, the patient did not respond well to this treatment, and radiographic assessments continued to show progression of the disease. Therefore, a fourth-line treatment of Atezolizumab in combination with Irinotecan and Surufatinib capsules was considered from October 27, 2021. After 2 cycles of treatment, CT evaluation showed that the efficacy achieved PR again ( Figure 1D ), blood examination showed a progressive decrease in AFP level to 12,238 ng/ml ( Figure 2A ). Considering this positive response, another 5 cycles of Atezolizumab in combination with Irinotecan and Surufatinib therapy were conducted with the last treatment on March 23, 2022, and the efficacy was assessed as sustained PR. The therapy was well tolerated by the patient during the whole treatment process, and the main adverse reactions were bone marrow suppression (leucopenia of 2 degrees) and digestive tract reactions (nausea and vomiting), which improved after treatment with Granulocyte-colony-Stimulating-Factor (G-CSF) and enhanced antiemetic therapy. There were no immune-related pneumonia, myocarditis, or other adverse effects observed. The patient currently has an overall survival of more than 14 months.
AFPGC is a special and rare subtype of gastric carcinoma. The histological diagnosis of AFPGC is more common in poorly differentiated adenocarcinoma, which is strongly associated with larger tumor volume, deeper serous membrane infiltration, and higher levels of invasion, lymph node, and liver metastasis (7). AFPGC has a poorer prognosis than AFP-negative gastric cancer. The median survival of advanced AFPGC is about 9.3 months (8). There is currently no standard treatment for this type of gastric cancer. However, early radical resection and active postoperative adjuvant chemotherapy have been shown to enhance the prognosis of AFPGC patients. For patients who have lost the chance of surgical treatment in the late stage, there is little literature on treatment. At present, the chemotherapy plan is mostly referred to as common gastric cancer, but the efficacy is worse than that of common gastric cancer. The median Overall Survival (OS) of patients is 9.3 months, and the 5-year survival rate is less than 20% (9). In 2018, Wang reported the efficacy of different chemotherapy regimens in 105 patients with advanced AFPGC, and the Overall Response Rate (ORR) of the platinum-based triple regimen was 56.1%, which was better than that of double-regimen (26.3%) (10). in 2021, Li reported that three patients received oral Apatinib 500 mg once a day combined with XELOX (Oxaliplatin/Capecitabine). One patient reached PR with progression-free survival(PFS) for more than 13 months in the first line treatment, and the other for 7 months, the third one had PFS for 3 months (11). Arakawa also reported that Ramucirumab targeting Vascular Endothelial Growth Factor Receptor 2 (VEGFR2) had some efficacy in AFPGC patients with an OS of 16 months (12). In this case, the patient was confirmed to have multiple liver metastases at the time of diagnosis and lost the opportunity of surgical treatment. The patient failed to benefit from the first-line two-drug combined chemotherapy, and did not respond to Apatinib, which confirmed that the therapeutic effect was worse than that of ordinary gastric cancer, and the patient’s treatment options were limited. Immunotherapy has demonstrated some efficacy in the treatment of advanced gastric cancer in recent years. The checkmate-649 study (13) showed that the combination of Nivolumab and chemotherapy increased the duration of PFS and OS in patients with Combined Positive Score (CPS) ≥ 5 and CPS ≥ 1 compared with chemotherapy alone, and a statistical difference was seen in the entire population (13.8 months versus 11.6 months, HR = 0.80). The results of ATTRACTION-4 showed that the median PFS time (10.5 months: 8.3 months, HR=0.68) and ORR (57.5% vs 47.8%, P=0.0088) were significantly better than those of chemotherapy alone. Furthermore, this was population-wide research with no molecular marker selected (14). The results of the Checkmate-649 and ATTRACTION-4 studies confirm the role of immunochemotherapy in the first-line treatment of gastric cancer. However, few studies have reported on the efficacy of advanced AFPGC immunotherapy. The patient described here, was a young female with MSS molecular typing and HER2 negative expression. The patient came with extensive liver metastasis and a significant tumor load. If the first therapy was ineffective, the patient might expect to live for only a short time. This patient did not benefit significantly from the three courses of first-line chemotherapy, and the AFP level was continuously above the critical value of 54,000 ng/ml. Due to the limited pathological tissue obtained by gastroscopy, the detection of PD-1 expression level could not be carried out. In the case of unknown PD-1 expression level, we chose Tiralizumab immunotherapy combined with oxaliplatin chemotherapy as the second line treatment. After 2 cycles of treatment, the AFP level decreased significantly and liver lesions shrank. PR was achieved after 4 cycles of treatment. The patient was relatively sensitive to PD-1 mAb combined with chemotherapy, but the remission period was short. Previous data suggest that Apatinib targeted anti-angiogenesis therapy was an effective way to overcome AFPGC. But unfortunately, this patient did not show efficacy in the treatment regimen of third-line. Sorafenib is one of the anti-VEGF drugs that has been reported to be effective (15, 16), but has not been reported in the treatment of AFPGC. In this case, the patient was changed to the PD-L1 ab therapy, when PD1 mAb resistance progressed. Surprisingly, it was found that the treatment effect was very good, and achieved immunotherapy re-challenge successfully. It demonstrated that switching from PD-1 inhibitor to PD-L1 inhibitor therapy may overcome potential resistance. The main mechanism is that PD-L1 mAb not only inhibit the PD1-PDL1 pathway, but also can activate DC cells and T cell functions by blocking the co-inhibition of B7.1 and PD-L1. At present, the disease is still in remission, there were no obvious adverse reactions during the whole treatment and it was well tolerated by the patient. The rise and fall in serum AFP levels during treatment were found to be positively correlated with the patient’s condition, and elevation of serum AFP level may be detected prior to appearance of symptoms and imaging detection. Therefore, measuring the serum AFP levels as a follow-up marker is an important means that can be used to evaluate condition changes of a patient. With the emergence of a new generation of gene sequencing technology, gastric cancer can be divided into different subtypes, based on gene mutations. In 2014, TCGA (The Cancer Genome Atlas) reported the results of genomic mapping of 295 cases of primary gastric Cancer and established four genomic subtypes (17), including microsatellite instability (MSI), Epstein-Barr virus infection (EBV+), and tumors with low aneuploidy (GS) genomic stability and high aneuploidy (CIN) chromosomal instability. According to Arora (18), loss of heterozygosity (LOH) is common in gastric cancer and can lead to chromosomal instability and the loss of tumor suppressor genes. The majority of tumors with increased AFP expression were categorized as chromosomal instability subtypes, with a 72% median index of allele loss. This was 50% higher than normal gastric adenocarcinoma. Patients with AFPGC may respond to immunotherapy due to their unique genetic characteristics. In addition, other studies demonstrated that CIN is a driver of metastatic progression, which may partially contribute to the aggressive phenotype of AFPGC. In all, the identification of these potential tumor drivers raises the potential for tumor-specific immunotherapy. Previous research reported that several molecular factors such as Vascular Endothelial Growth Factor-C (VEGF-C), Signal Transducer and Activator of transcription 3 (STAT3), and Hepatocyte Growth Factor (HGF) seem to over-expressed more frequently in AFPGC than in stage-matched non-AFPGC (5, 19). Chen (20) found that ANGPTL6 is an important driver gene of angiogenesis in AFPGC development. ANGPTL6 promotes endothelial cell migration and tube formation through activation of ERK1/2 and AKT pathways. ANGPTL6 knockdown inhibits cancer cell apoptosis and invasiveness. These findings provide not only effective biomarkers for diagnosis but also attractive therapeutic targets for AFPGC patients. In conclusion, AFPGC, as a special type of gastric cancer, has a low incidence but a high degree of malignancy. Improving the understanding of this type of gastric cancer can avoid misdiagnosing it as common gastric adenocarcinoma and underestimating its malignancy. At present, only a few cases have been reported on the conversion of PD-L1 mAb after the progress of PD-1 mAb therapy, and there is still a lack of large-scale prospective studies. However, there are still many open questions, for example, how to choose the treatment plan after the progress of PD-L1 mAb therapy? How to optimize the immune combination therapy model? Despite the need of further studies to tackle those questions, this case report represents an important exploration and potential breakthrough in the treatment of advanced gastric cancer with immunotherapy.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
The studies involving human participants were reviewed and approved by the ethics committee of Suzhou Municipal Hospital. The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Conceptualization: FG, JS and TZ. Attending physicians for the patient: YF and LZ. Writing—original draft: LW. Editing draft: AH and BL. Supervision, FG, JS, and QZ. All authors contributed to the article and approved the submitted version.
This work was supported by grants from Special Project of Diagnosis and Treatment Technology of Key Clinical Diseases of Suzhou City (LCZX201910) and the Medical Scientific Research Project of Jiangsu Commission of Health (Grant No.Z2020003 and Z2020058). The funders of the current study had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647059 | Ling Zhao,Chen Hou,Naihong Yan | Neuroinflammation in retinitis pigmentosa: Therapies targeting the innate immune system | 27-10-2022 | retinal inflammation,innate immune,gut microbiome,trained immunity,epigenetic modification,retinitis pigmentosa | Retinitis pigmentosa (RP) is an important cause of irreversible blindness worldwide and lacks effective treatment strategies. Although mutations are the primary cause of RP, research over the past decades has shown that neuroinflammation is an important cause of RP progression. Due to the abnormal activation of immunity, continuous sterile inflammation results in neuron loss and structural destruction. Therapies targeting inflammation have shown their potential to attenuate photoreceptor degeneration in preclinical models. Regardless of variations in genetic background, inflammatory modulation is emerging as an important role in the treatment of RP. We summarize the evidence for the role of inflammation in RP and mention therapeutic strategies where available, focusing on the modulation of innate immune signals, including TNFα signaling, TLR signaling, NLRP3 inflammasome activation, chemokine signaling and JAK/STAT signaling. In addition, we describe epigenetic regulation, the gut microbiome and herbal agents as prospective treatment strategies for RP in recent advances. | Neuroinflammation in retinitis pigmentosa: Therapies targeting the innate immune system
Retinitis pigmentosa (RP) is an important cause of irreversible blindness worldwide and lacks effective treatment strategies. Although mutations are the primary cause of RP, research over the past decades has shown that neuroinflammation is an important cause of RP progression. Due to the abnormal activation of immunity, continuous sterile inflammation results in neuron loss and structural destruction. Therapies targeting inflammation have shown their potential to attenuate photoreceptor degeneration in preclinical models. Regardless of variations in genetic background, inflammatory modulation is emerging as an important role in the treatment of RP. We summarize the evidence for the role of inflammation in RP and mention therapeutic strategies where available, focusing on the modulation of innate immune signals, including TNFα signaling, TLR signaling, NLRP3 inflammasome activation, chemokine signaling and JAK/STAT signaling. In addition, we describe epigenetic regulation, the gut microbiome and herbal agents as prospective treatment strategies for RP in recent advances.
Retinitis pigmentosa (RP) is a category of inherited retinal dystrophies marked by vision loss and, ultimately, blindness. More than 3000 mutations in over 80 distinct genes or loci have been identified as causes of non-syndromic RP (1, 2). These mutations can be transmitted in an autosomal-dominant, autosomal-recessive, or X-linked manner. There are also syndromic forms of RP, such as Usher syndrome and Bardet-Biedl syndrome (3). The prevalence of RP is reported to be 1/3,000 to 1/5,000 (https://www.orpha.net/consor/cgi-bin/index.php?lng=EN). Early in the progression of RP, patients experience night blindness and difficulty with dark adaptation. They gradually lose peripheral vision and develop tunnel vision as the disease advances, indicating the loss of rod function. Cone involvement contributes to visual acuity decline over time, and finally, typically in middle age, central vision loss occurs (2). Due to its clinical and genetic heterogeneity, RP has limited therapeutic options. It has long been recognized that inflammation and immune responses are associated with RP, and this theme has recently gained increasing attention (4–6). Greater understanding of the molecular processes driving RP inflammation is expected to provide new therapeutic approaches independent of the genetic background. Inflammation activation is a prominent feature of RP. In RP, inflammation is characterized by activation of the innate immune system, including dysfunction of the immune barrier, activation and infiltration of immune cells, and upregulation of topical and peripheral inflammatory factors. Bone-spicule pigmentation, attenuated retinal vessels and waxy pallor of the optic disc are typical clinical manifestations of RP. In addition, inflammatory cells are commonly observed in the vitreous due to the collapse of the blood−retina barrier (BRB). Higher cell density correlates with younger age and impaired visual function (6). In addition, it has been reported that increased aqueous flare in RP patients correlates closely with visual function and the extent of global retinal degeneration (7–11). Aqueous flare is generally seen in individuals with inflammatory ocular disorders, indicating deficits of the blood–aqueous barrier and inflammatory protein/cell leakage (12, 13). BRB disruption begins early in the disease. Prior to the infiltration of inflammatory cells and photoreceptor (PR) starvation, the tight junctions of the retinal pigment epithelium (RPE) and the retinal vasculature become leaky, thereby promoting the formation of an inflammatory milieu and the degeneration of PRs (14–17). Microglia are essential components of the retinal innate immune system and play a pivotal role in retinal inflammatory responses. Gupta et al. reported that microglial activation is engaged in human RP. With thinning of the PR layer, microglia were observed to infiltrate degenerative foci; these microglia were enlarged amoeboid cells containing rhodopsin-positive cytoplasmic inclusions (18). Microglia promote retinal inflammation via infiltration, phagocytosis, and secretion of proinflammatory mediators, whereas genetic ablation or inhibition of microglial phagocytosis ameliorates PR degeneration in RP model mice (19). Several studies (6, 20–22) have reported elevated levels of inflammatory factors in serum, vitreous, and aqueous humor, indicating a proactive inflammatory response in RP patients. Immunologic disorders observed in patients with RP are summarized in a previous work (23). Animal models are essential for the study of RP (24, 25). Activation of the immune system has been detected in various rodent RP animal models (26–28). Due to this similarity, animal models are indispensable for clarifying the pathogenesis of RP and developing treatments. Clinically, synthetic corticosteroids with potent anti-inflammatory and immunosuppressive properties are used in treatments for RP-related cystoid macular edema (29). In RP models, it also works. In RCS and rhodopsin mutant model S334ter-4 rats, fluocinolone acetonide treatment markedly protects PR from degeneration and suppresses microglial activity (30, 31). In combination with polyamidoamine dendrimers, fluocinolone acetonide selectively targets the microglia localized in the outer retina where degeneration is ongoing (32). Dexamethasone administration to rd10 mice reduces retinal inflammation, restores cone structure and function, and preserves RPE integrity by preserving ZO-1 density (15, 33).
Microglial activation is a sign of neuroinflammation. Retinal resident microglia arise from yolk sac erythromyeloid progenitors, comprising 85% of total retina macrophages (34–36). They colonize the developing retina during embryogenesis, shape the retina by secreting neurotrophic factors, engulf and eliminate unwanted neurons and synapses, and engage in vascular development of the eye (37–39). In postnatal retinas, microglia are maintained throughout life independent of circulating monocytes and by self-renewal (34, 40, 41). Microglia express a variety of receptors (e.g., CX3CR1, TLRs, IL-1R, and TNFR) that allow them to detect environmental changes and initiate the inflammatory signal cascade (42). Microglia activation induces a robust inflammatory response, including the release of proinflammatory factors, phagocytosis, and inflammatory cell recruitment. Excessive microglial phagocytosis contributes to local inflammation and neurodegeneration (43, 44). Classically, activated microglia were categorized into two groups: M1 (classically activated) and M2 (alternatively activated), with the belief that M1 microglia secrete proinflammatory factors such as TNFα, IL-1β, IL-6, and inducible nitric oxide synthase (iNOS) that fuel inflammation, whereas M2 microglia produce anti-inflammatory cytokines (e.g., IL-4, IL-10, IL-13, IL-18) that are beneficial for damage repair (45). Recent research, however, suggests that the microglial phenotype varies in response to environmental changes (46). In healthy retinas, microglia tile the inner and outer plexiform layers without overlapping (37), where they are ramified cells responsible for immune surveillance and maintenance of synaptic structure and transmission (47–50). In response to insults, microglia rapidly change into an amoeboid appearance and migrate into lesion areas, removing dead/dying neurons and neuronal debris while concurrently releasing proinflammatory factors as well as protective cytokines and trophic factors to repair damage and restore homeostasis (51), after which microglia recover to the “resting” state; this process usually results in minimal retinal remodeling. In RP retinas, initial mutation-driven PR degeneration increases extracellular signal molecules (e.g., ATP, HSPs, HMGB1, DNA, and many others) termed damage-associated molecular patterns (DAMPs) (52–56). The “eat-me” signal, phosphatidylserine, appears on stressed rods (19). Microglia proliferate and infiltrate the PR layer and subretinal space, where they function as reactive phagocytes, phagocytose dead and stressed PRs, secrete proinflammatory cytokines (e.g., TNFα and IL-1β) and chemokines (e.g., CCL2 and RANTES), and recruit infiltrating immune cells (19, 24, 57, 58). Due to this mutant genetic background, however, microglial activation persists, and the continuous production of inflammatory and cytotoxic factors exacerbates PR loss until the late stage, at which point PRs mostly die and the retinal structure is severely damaged (59). Müller glia are another group of retinal cells engaged in degeneration. Müller glia are retinal macroglia that provide homeostasis, metabolism, and functional support for neurons (60). Depending on the severity, the Müller glial response to injury refers to reactive gliosis accompanied by Müller proliferation or not. Reactive gliosis can be beneficial because it releases protective factors such as neurotrophic factors, whereas prolonged gliosis is detrimental and generally results in neurodegeneration (61). Müller glia are potential modulators of retinal inflammation. Upon BRB disruption, Müller glia compensate for RPE deficiency by sealing the leaky choroid and inducing claudin-5 expression (14). Müller glia share characteristics with immune cells. Müller glia express multiple cytokine receptors and are a major source of cytokines and inflammatory factors (62). Proteomic evidence supports the capacity of Müller glia for antigen presentation and inflammatory signaling transduction in response to immune stimulation (63, 64). Müller glia contribute to the phagocytic clearance of dead PRs (65). Moreover, the interaction between Müller glia and microglia modulates retinal inflammation and degeneration (66–68). As the predominant glial population of the retina, Müller glia are abundant and widely distributed. Müller glia traverse the thickness of the neuroretina structurally, allowing them to keep touch with all types of retinal cells. Due to its neurotrophic function and regeneration potential, the Müller cell has been studied in a variety of degenerative retinal disorders (69). In actuality, Müller glia are also intimately linked to retinal inflammation. For more information about how Müller glia interact with the innate immune system and monitor retinal inflammation, we refer the reader to this article (70). Complicated mechanisms are involved in the regulation of retinal inflammation in RP. Microglia and Müller glia are major cellular populations that express and modulate these signaling pathways (Figure 1). Here, we review treatment strategies from the perspective of inflammation management (Table 1), focus on molecular mechanisms related to immunomodulation, and discuss new findings regarding epigenetic modification and the gut microbiome as novel therapies for RP.
Tumor necrosis factor α (TNFα) is a strong proinflammatory cytokine that plays vital roles in immune modulation, cell proliferation, differentiation, and apoptosis. TNFα is produced predominantly by T and innate immune cells and is initially synthesized as transmembrane protein (tmTNFα), a precursor that requires proteolytic cleavage by TNFα-converting enzyme (also ADAM17) to release a soluble form (sTNFα) (124). Both tmTNFα and sTNFα are implicated in the inflammatory response. TNFα initiates a signal cascade by binding to its receptors, TNFR1 and TNFR2. TNFR1 is activated by both tmTNFα and sTNFα, whereas TNFR2 is proposed to be fully activated primarily by tmTNFα (125). Ligand binding to TNFR1 recruits the adaptor molecule TNFR1-associated death domain protein, which then leads to the assembly of several signaling complexes known as complexes I, IIa, IIb, and IIc. Complex I formation stimulates nuclear factor kappa B (NF-κB) and mitogen-activated protein kinases (MAPKs). Complex IIa and IIb assembly activates caspase-8 and facilitates apoptosis, and complex IIc formation activates the mixed lineage kinase domain-like protein and induces necroptosis and inflammation. TNFR2 stimulation activates NF-κB, MAPKs, and protein kinase B (125). TNFα is postulated to participate in the pathogenesis of RP (74). TNFα and TNFR expression levels are elevated in the retina of RP models and in the aqueous humor of RP patients (24, 74, 126–128); microglia (129) and Müller glia (130, 131) are the primary cellular sources of TNFα. TNFα signaling has been found to mediate PR death via RIP1/3-related necrosis and caspase3/7-dependent apoptosis, in addition to triggering proinflammatory signaling in the RP retina (73, 126).
Increased TNFα expression in the retina is linked to nerve growth factor (NGF) receptor activation. Müller glia in rhodopsin mutant RP model RHOP347S mice upregulate the expression of TrkC. T1, a truncated TrkC receptor isoform, and its ligand NT-3. TrkC.T1 increases local TNFα production by activating MAPK/Erk, ultimately leading to PR death. This process can be reversed by genetic knockdown of TrkC.T1, TrkC antagonism, or MAPK/Erk inhibition (71). Similarly, TrkC.T1 knockout (KO) and TrkC inhibition increased retinal ganglion cell survival in a mouse model of glaucoma by reducing TNFα production (132), implying that TrkC.T1 is upstream of TNFα. It has been reported that microglia-derived proNGF facilitates PR death via p75NTR (133), proNGF binding to p75NTR in Müller glia induces robust expression of TNFα and TNFα-dependent neuron death in rodent retina (131, 134), the expression levels of proNGF and p75NTR are increased in the retina of rd10 at early degenerative stages, pharmacological antagonism of p75NTR with THX-B ((1,3-diisopropyl-1-[2-(1,3-dimethyl-2,6-dioxo-1,2,3,6tetrahydro-purin-7-yl)-acetyl]-urea)) affords neuroprotection to PRs, and the treatment also mediates reduction of TNFα production, microglial activation, and reactive gliosis (72).
TNFα knockdown in the T17M rhodopsin mutant mouse model reduces PR death and PR-related functional loss, and this neuroprotective effect is associated with reductions in proinflammatory cytokines (IL-1β, IL-6, IL-17, RANTES) and chemokines (CCL2) (73). Infliximab and adalimumab are biological TNFα inhibitors approved for treating inflammatory disorders such as Crohn’s disease, ulcerative colitis, rheumatoid arthritis, plaque psoriasis, and uveitis (125, 135). By lowering the expression of TNFR1 and caspase-3 activity, infliximab alleviated retinal degeneration induced by PDE6 inhibition in cultured porcine retina (74, 127). Adalimumab administered intraperitoneally or topically improved PR survival while decreasing microglial activation and reactive gliosis in the rd10 retina. Inhibition of PARP and RIPK signaling, as well as NLRP3 inflammasome assembly, are mechanisms involved in this protective response (25, 75).
Notably, TNFα KO retinas tend to express both pro- and anti-inflammatory factors at reduced levels when compared with controls (73), implying that removing TNFα would also damage the immune system’s defenses. Recent work by Kuhn et al. established that TNFα, in collaboration with TNFR1, TNFR2, and p75NTR, induces signals that are indispensable for neural development and that disturbances to TNFR family signaling result in unhealthy axonal development (136). ADAM17 regulates the expression of TNFα as well as the receptor TNFR (124). Muliyil et al. reported that ADAM17 and soluble TNF mediate a novel cytoprotective pathway in Drosophila. Loss of ADAM17 or TNF/TNFR signaling drives the accumulation of lipid droplets and degeneration in the Drosophila retina, whereas restoration of ADAM17 or TNF/TNFR in glia is sufficient to rescue the degeneration phenotype. TNF and TNFR are explicitly needed in glia; loss of either in glia, but not neurons, leads to the accumulation of lipid droplets. Furthermore, inactivation of ADAM17 in human iPSC-derived microglia similarly induces aberrant lipid droplet accumulation and mitochondrial reactive oxygen species generation (137), indicating that comparable processes in which TNF works not as an inflammatory trigger but as a trophic survival factor (137) may also be involved in the mammalian retina. Benoot et al. evaluated the numerous contradictory findings of TNFα application in lung cancer (138), bringing to our awareness the varied functions of various TNF family members and the positive impacts of TNF signaling. Modern genomic, transcriptomic, and proteomic techniques are useful for identifying signaling events and molecules in signal transduction (139, 140). Tanzer et al. (141) discussed in detail how modern proteomic approaches offer a novel perspective on TNF signaling. Currently, the precise mechanisms of TNFα synthesis remain obscure. Future investigation of TNFα signaling requires the power of new technology, and the potential protective function of TNFα merits greater consideration.
Müller glia exposed to activated microglia modify the expression of a variety of signaling molecules, including (1) elevation of growth factors such as GDNF and leukemia inhibitory factor (LIF), (2) enhanced proinflammatory factor production, and (3) overexpression of chemokines and adhesion proteins (142). TNFα is the most prevalent cytokine produced by reactive microglia, and it stimulates LIF expression in Müller glia in a p38MAPK-dependent manner. Inhibition of p38 MAPK activity lowered LIF expression and accelerated PR mortality in light-damaged retinas (143), similar to previous reports that TNFα prevents cell death by activating the JAK/STAT3 pathway through the IL-6 receptor (144). When TNFα stimuli engage previously activated Müller glia, however, inducible cytokines consisting of more proinflammatory cytokines (TNFα, iNOS, IL-6) and less LIF are produced (145). In other words, depending on the type and degree of stimuli, Müller glia activation generates both neuroprotective and proinflammatory responses, and Müller glia under continuous stimulation are likely to exhibit a detrimental phenotype.
Toll-like receptors (TLRs) are a class of pattern recognition receptors (PRRs) responsible for identifying pathogen-associated molecular patterns (PAMPs) and DAMPs and mediating immune responses; the generation of PAMPs or DAMPs prompts pathogen invasion or tissue injury. TLR expression is conserved among species, and to date, 10 TLRs (TLR1–10) in humans and 12 TLRs (TLR1–9 and TLR11–13) in mice have been described. TLRs are predominantly but not exclusively expressed on immune cells (146–148). TLRs serve as the first line of defense for the innate immune system. Upon recognition of DAMPs or PAMPs, TLRs dimerize and initiate recruitment of Toll/IL-1 receptor (TIR) domain-containing adaptor molecules, including myeloid differentiation primary response 88 (MyD88), TIR-domain-containing adaptor-inducing interferon-β (TRIF), MyD88 adaptor-like protein (Mal), and TRIF-related adaptor molecule (TRAM), thereby initiating intracellular signaling cascades: the MyD88- or TRIF-dependent pathways (146). TLR activation facilitates the transduction of NF-κB and MAPK (149–151), as well as the release of proinflammatory cytokines (TNFα, IL-6, IL-1β, and IFNβ), chemokines, and cluster of differentiation 80 (CD80), CD86, CD40, and major histocompatibility complex class II (146). Activation of TLR signaling has been shown to worsen inflammation and accelerate the course of RP (77, 152, 153). Microglia highly express TLRs (154), and TLR activation in the retina facilitates microglial activation and infiltration (77, 155). Moreover, microglia in the rd1 retina undergo RIP1/RIP3-dependent necroptosis mediated by TLR4 activation, which amplifies retinal inflammation and destruction with large amounts of proinflammatory cytokines (TNFα and CCL2) (152).
High-mobility group box-1 (HMGB1) is a proinflammatory factor and DAMP released by dying cells or activated macrophages that mediates the immune response via PRRs (156–158). HMGB1 stimulates an inflammatory response in diabetic retinopathy through TLR4/NF-κB signaling (159). Increased levels of HMGB1 were detected in the vitreous of patients with RP, along with the presence of necrotic enlarged cone cells (53). In cultured cone-like 661W cells, recombinant HMGB1 treatment induces apoptosis and upregulates the expression of IL-6 and TNFα (160), and external HMGB1 induces retinal ganglion cell death via TLR2/4 signaling (161), whereas HMGB1 inhibition or neutralization attenuates the inflammatory response and promotes retinal neuron survival (162, 163).
Upregulation of Tlr2, Il1b, Myd88 and Tirap was found in RP model rd10 and P23H mice, demonstrating TLR activation involvement in RP-associated retinal degeneration. Genetic deletion of TLR2 alleviated PR loss and vision impairment in both models (76). Similarly, in a light-induced retinal degeneration model, genetic TLR4 deletion reduced retinal inflammation and degeneration (77). Minocycline is an effective microglial inhibitor. In inherited and induced RP models, minocycline administration decreased microglial infiltration and proinflammatory molecule expression and promoted PR survival and functional retention (78–82). Minocycline treatment suppresses MAPK and NF-κB signaling in LPS-stimulated microglia (164), and it has been ascertained that minocycline prevents microglial activation by inhibiting TLR2 (165, 166) and TLR4 (167) signaling.
Most TLRs (except for TLR3) use MyD88 as a downstream adaptor protein; moreover, MyD88 is a component of the IL-1R signaling cascade (168, 169). MyD88 features a death domain and a TIR domain. Upon TLR/IL-1R ligation, MyD88 is recruited to the receptor and interacts with IRAK2/4 through their death domains, which activates NF-κB, activator protein-1, and interferon regulatory factors (169). MyD88 KO mice display attenuated immune responses and are unable to produce normal levels of inflammatory cytokines (170). This diminished immune response preserved PR survival and retinal function during degeneration in rd1 mice lacking MyD88 (83). Similarly, pharmacologic inhibition of MyD88 in rd10 mice with MyD88 inhibitor peptide reduced PR apoptosis and improved rod-related function; treatment also lowered the number of microglia in the PR layer and increased microglia/macrophage expression of the neuroprotective marker Arg1 (84). Further proteomic analysis demonstrated that treatment with such MyD88 inhibitor peptides boosted crystalline expression, suggesting that MyD88 inhibition may also enhance intrinsic tissue-protective mechanisms (85).
Activated microglia/macrophage WAP domain protein (AMWAP), secreted by reactive microglia, is a hallmark of microglial activation. While AMWAP overexpression in microglia lowers the production of proinflammatory factors such as IL-6, iNOS, CCL2, CASP11, and TNFα, extracellular AMWAP endocytosed by microglia inhibits TLR2- and TLR4-induced NF-κB translocation by preventing IRAK-1 and IκBα proteolysis (86). AMWAP administration lowers the apoptosis of 661w cone-like cells treated with microglia-conditioned medium (86), indicating that AMWAP is a potential self-modulator of TLR signaling in microglia. The TLR signaling pathway plays a fundamental role in inflammatory and immune responses. Molecules released from injured neurons induce an intracellular signaling cascade through TLR/MyD88, contributing to further retinal damage. Blockage of TLR/MyD88 alleviates RP by reducing inflammatory responses and enhancing protective effects.
Inflammasomes are cytosolic multiprotein complexes that facilitate the release of mature IL-1β, IL-18, and cleaved caspase-1. The intracellular PRRs, NOD-like receptors (NLRs), are important components of the inflammasome complex. Some NLRs oligomerize upon activation to form multiprotein complexes that function as caspase-1-activating scaffolds (171). NLRP3 is the most well-studied NLR; NLRP3 inflammasome assembly requires two signals: a priming signal that activates NF-κB, followed by transcription of NLRP3, pro-IL-1β, and pro-IL-18. A second activation signal facilitates the recruitment and oligomerization of NLRP3, adaptor protein ASC (apoptosis-associated speck-like protein containing a CARD), and pro-caspase-1. Once the inflammasome is assembled, it stimulates pro-caspase-1 self-cleavage and activation, and cleaved caspase-1 catalyzes pro-IL-1β and pro-IL-18 maturation and induces the release of their mature forms. The recognition of DAMPs or PAMPs that act through PRRs such as TLRs or cytokines that act through particular receptors (TNFR, IL-1R) exemplifies the priming signal. The activation signal encompasses a wide range of stimuli, including ion flux (K+, Cl-, Ca2+), lysosomal instability, mitochondrial dysfunction, reactive oxygen species generation, and trans-Golgi disassembly, with K+ efflux being the upstream event in almost all NLRP3 activations (172, 173). Inflammasome activation initiates the host’s defense response to endogenous or external damaging stimuli and aids in homeostasis maintenance. Nevertheless, chronic inflammasome activation and the subsequent overproduction of caspase-1, IL-1β, and IL-18 can be detrimental. Canine models of RP upregulate NLRP3 inflammasome-related genes (26). NLRP3 was detected in cone PRs and one-third of reactive microglia in P23H rhodopsin mutant retinas, which also upregulates the expression of mature IL-1β and IL-18, as well as cleaved caspase-1, indicating inflammasome activation during retinal degeneration. In rd10 mice, administration of the antioxidant N-acetylcysteine prevented PR loss and suppressed inflammatory factors and microglial activation (24). Studies conducted on P23H mice demonstrated that N-acetylcysteine lowered NLRP3 expression by 50% and decreased microglial infiltration, hence improving cone survival and retinal function (87).
The purinergic receptor P2X7R is an adenosine triphosphate (ATP)-gated ion channel and a well-known inflammasome activator that can enhance the expression of the NLRP3 inflammasome in microglia (174). By inducing K+ efflux, ATP-mediated P2X7R activation promotes NLRP3 inflammasome activation (173). ATP is abundant in PRs as an energy source and neurotransmitter. During retinal degeneration, ATP leaches from dying PRs and activates P2X7R (175). Intravitreal injection of PPADS (pyridoxal-phosphate-6-azophenyl-2’,4’-disulfonic acid), a purinergic antagonist, lowers PR loss in rd1 mice (88). In contrast, intravitreal administration of ATP to WT (wild-type) retinas induces PR degeneration similar to that in the P23H RP model (176), whereas treatment with the selective P2X7R inhibitor BBG (Brilliant Blue G) protects against this ATP-mediated PR apoptosis (177). BBG therapy also reduced inflammasome components (NLRP3, cleaved caspase-1 and mature IL-1β proteins) in P23H retinas (87). In the absence of extracellular ATP, P2X7R functions as a scavenger receptor that governs microglial clearance of extracellular debris, whereas P2X7R overactivation triggers NLRP3 inflammasome activation by provoking lysosomal instability. Lowering extracellular ATP levels may have the dual benefit of enhancing phagocytosis while decreasing inflammation (178).
IL-1β is a key product of NLRP3 inflammasome activation and a potent immunomodulation factor that orchestrates inflammatory and host defense responses (172, 179). IL-1β signals through IL-1R1. IL-1β binding to IL-1R1 stimulates pathways such as NF-κB, p38, JNKs, ERKs, and MAPKs, facilitating inflammatory cell recruitment and local/systemic inflammatory responses (180). Appropriate IL-1β/IL-1R1 signaling is required for a host’s defensive response to infections, whereas excessive IL-1β signaling is seen in a variety of hereditary and nonhereditary autoinflammatory disorders. IL-1β activity is endogenously regulated by IL-1R2 and IL-1Ra; IL-1R2 is a decoy receptor that sequesters the IL-1β signal, while IL-1Ra blocks IL-1β by competitively binding to IL-1R1 (180). Intravitreal delivery of exogenous IL-1β triggered an immediate inflammatory response in the retina, including leukocyte recruitment and BRB destruction (181). However, IL-1β does not trigger PR death directly, as IL-1R1 expression is low in PRs. Through IL-1R1 expressed on Müller glia, IL-1β drives glutamate excitotoxicity-induced rod PR loss. The IL-1β/IL-1R1 signal disrupts the process of glutamate conversion into glutamine in Müller glia, resulting in an increased intracellular glutamate concentration, and upregulates xCT (the core subunit of the cystine/glutamate transporter system xc-) expression, which facilitates glutamate release into the extracellular space. Furthermore, IL-1β upregulates the expression of the ionotropic glutamate receptor in retinal neurons, which may increase neuronal vulnerability to glutamate excitotoxicity (182). Infiltrating microglia in the rd10 retina upregulate the expression of IL-1β (19) and block IL-1β signaling using anakinra, a commercially available recombinant IL-1Ra that is fully active in blocking IL-1R1 (183), which reduces PR apoptosis and preserves outer nuclear layer thickness in rd10 animals (19). In contrast, Todd et al. (184) demonstrated that IL-1β expressed by reactive microglia provides neuroprotection via IL-1R1 expressed on astrocytes in another mouse model of NMDA-induced retinal degeneration. Despite the use of different models, we were able to determine that IL-1β acts on the surface receptors of distinct glial cells and has varying effects on PR survival.
CX3CR1 expression in the central nervous system (CNS) is considered to be restricted to microglia, and the expression of its sole ligand, CX3CL1 (also known as fractalkine), is confined to certain neurons (185). CX3CL1/CX3CR1 signaling facilitates the interaction between neurons and glia and plays a vital role in CNS neuroinflammation (186, 187). CX3CL1/CX3CR1 signaling contributes to normal microglial and PR function. CX3CR1 signaling governs the dynamic activity of retinal microglia (188). Microglial ablation and repopulation in the mouse retina have shown that microglial recruitment is regulated by CX3CL1/CX3CR1 signaling (40), and Müller glia augment microglial migration and infiltration by increasing CX3CL1 secretion and microglial CX3CR1 expression (68). In addition, CX3CR1 signaling is required for retinal neuron growth, as CX3CR1-deficient retinas have shorter outer segments and diminished cone-related retinal function (189). CX3CL1/CX3CR1 signaling affects microglial homeostasis by modulating the inflammatory response and phagocytosis. Increasing CX3CL1/CX3CR1 signaling in RP retinas could be beneficial. CX3CR1 deficiency impairs microglial phagocytic clearance of neurotoxic species. Reportedly, CX3CL1 signaling enhances microglial erythrophagocytosis through the CD163/HO-1 axis (190), whereas CX3CR1 KO weakens microglial phagocytosis to β-amyloid and mediates lysosomal dysfunction, resulting in an escalation of neuroinflammation due to β-amyloid accumulation (191). CX3CR1 deficiency enhances the inflammatory response of microglia. CX3CR1-deficient microglia exhibit greater neurotoxicity (192), and CX3CR1-deficient microglia have an elevated amount of surface P2X7R, which increases IL-1β maturation and release (193). CX3CR1 deletion in microglia-like cells generated from human iPSCs induced enhanced inflammatory responses to LPS stimuli and phagocytic activity to fluorescent beads (194). Loss of CX3CR1 signaling in young animals resulted in a microglial transcriptome similar to that of aged mice, with dysregulated expression of genes related to immune function (195). CX3CL1 expression is downregulated in rd10 retina before the onset of primary rod degeneration (196), and CX3CR1 KO in rd10 mice increases microglial infiltration and phagocytosis, as well as the generation of pro-inflammatory cytokines, which accelerates PR loss (82, 89), whereas exogenous CX3CL1 supplementation preserves morphology and function (89). CX3CL1 has been shown to deactivate microglia by blocking the NF-κB pathway and activating the Nrf2 pathway (197). A norgestrel-supplemented diet protected rd10 retinas from PR degeneration, and this protection was achieved by the upregulation of CX3CL1/CX3CR1 signaling and the reduction of proinflammatory cytokine production (91, 92). Recent work by Wang et al. demonstrated that overexpression of soluble CX3CL1 via AAV8 prolongs cone survival and improves cone-related visual function in RP model rd1 and rd10 mice. This therapeutic effect is restricted to cone PRs, has no effect on microglial activity or inflammatory factor levels and is not even dependent on the presence of a normal number of microglia (90). In light of this, further research is needed to determine whether CX3CL1 action in the retina is limited to microglia or whether other pathways exist.
Part of the evidence suggests that CCL2/CCR2 signaling is detrimental, as inhibition of CCL2/CCR2 signaling attenuates microglial activity and degeneration in RP (95–97). CCL2 is highly expressed by stressed PRs, activated microglia, and Müller glia in degenerating retina (95, 198, 199). By binding to its receptor, CCR2, which is expressed on peripheral mononuclear phagocytes, mediates the influx of circulating monocytes into inflamed retinas (200). Using fluorescent protein-labeled Mertk (-/-) Cx3cr1 (GFP/+) Ccr2 (RFP/+) mice, Kohno H and colleagues demonstrated that both minocycline and lecithin-bound iodine (LBI) ameliorate PR death by inhibiting CCL2/CCR2 signaling (93, 94). Meanwhile, constitutive expression of CX3CR1 in the retina represses CCL2 expression and the recruitment of neurotoxic inflammatory CCR2+ monocytes (201). However, there is evidence that CCL2 signaling may have a protective role in the degradation of RP. In a light-induced mouse model of degeneration, blocking CCL2/CCR2 signaling decreased infiltrating monocytes but had no effect on the rate of retinal thinning (198). Alde-Low EPCs (low aldehyde dehydrogenase activity endothelial progenitor cells) transplantation therapy rescued vasculature and PRs in rd1 mice, and CCL2 secreted by Alde-Low EPCs recruited a subpopulation of monocyte-derived macrophages that highly expressed CCR2 and the neuroprotective factors TGF-β, IGF-1 and IL-10 (202). In brief, induction of CCL2 expression by Alde-Low EPCs in rd1 retinas resulted in the recruitment of neuroprotective macrophages. It is apparent that CCL2/CCR2 signaling mediates the recruitment of monocyte-derived macrophages in the degenerating retina, but it remains to be determined whether these recruited cells are beneficial or detrimental.
The JAK/STAT signaling pathway is a ubiquitously expressed intracellular signal transduction system implicated in a wide range of biological functions. Various ligands, including cytokines, growth hormones, growth factors, and their receptors, can activate the JAK/STAT pathway (203). Briefly, ligand binding to specific receptors induces receptor multimerization and JAK activation, activated JAKs phosphorylate the receptors, activate and phosphorylate their primary substrate STAT, and phosphorylated STAT dimerizes and translocates into the nucleus, where it binds to particular regions to either activate or inhibit the transcription of target genes. Suppressor of cytokine signaling (SOCS) is a negative modulator of JAK/STAT signaling, and its expression is promoted by stimulation of JAK/STAT signaling (203, 204). Numerous studies on the JAK/STAT pathway have revealed its significance in neoplastic and inflammatory disorders (203, 205).
Expression and activation of STAT proteins are implicated in the plasticity of the retina during embryonic and postnatal stages (206), and mice deficient in SOCS1/STAT1 develop severe ocular illnesses with massive inflammatory cell infiltration (207). STAT signaling plays a central role in the degeneration of the rd10 retina, as evidenced by proteomic profiling (208). Furthermore, activation of JAK/STAT signaling was also observed in the retinas of light-induced and inherited (rd1 and VPP mouse) RP animal models (98, 209). AG490 is a JAK2-specific inhibitor that suppresses microglial activation and the production of inflammatory factors such as TNFα and IL-6 by reducing STAT3 phosphorylation (210). AG490 induces M2-type microglial polarization by blocking JAK2/STAT3 signaling in acute paraquat exposure-induced microglial activation (211). In light-damaged retinas, AG490 treatment decreased JAK and STAT phosphorylation as well as PR apoptosis (98). Olfactory ensheathing cell (OEC) transplantation improved retinal function in RCS rats. OEC treatment dramatically reduced active resident microglia/infiltrated macrophages and the release of proinflammatory cytokines while increasing anti-inflammatory cytokines in the transplantation area. This neuroprotection appears to be mediated in part by increased SOCS3 expression and decreased JAK2/STAT3 activity. Coculture of OECs with the BV2 microglial cell line revealed a shift in microglial cytokine release toward an anti-inflammatory pattern (99). According to the literature, SOCS3-deficient microglia display increased phagocytic activity (212), whereas elevated SOCS3 expression in microglia decreases GM-CSF/IFN-γ-driven inflammatory responses by blocking the activities of JAK1 and JAK2 through its KIR domain (213). In addition, increasing SOCS1 signaling with SOCS1-KIR, a SOCS1 mimetic peptide, suppressed the recruitment of inflammatory cells into the retina and stimulated IL-10 production (214). Multiple jakinibs (JAK inhibitors) are approved for the clinical management of malignancy, rheumatic, lymphoproliferative, and inflammatory diseases, and most recently, coronavirus disease 2019 (205), but their efficacy in the treatment of RP has not been evaluated.
Activation of JAK/STAT signaling has a protective effect on the RP retina. pMSC-derived retinal progenitor cell transplantation increased PR preservation in rd12 mice, and this protection was partially mediated by activation of the JAK/STAT pathway (100). Application of ciliary neurotrophic factor (CNTF) in RP preclinical research has gained significant neuroprotection and has been employed in clinical trials (101, 215, 216).. CNTF therapy enhances the protective properties of Müller glia through LIF/gp130/STAT3 signaling, thereby preventing retinal degeneration. CNTF treatment elevates the expression of LIF and endothelin 2 (Edn2) (102), and LIF is essential for CNTF-elicited STAT3 activation (217). LIF belongs to the IL-6 cytokine family and signals through the gp130 receptor. In a mouse model of light-induced retinal degeneration, intravitreal delivery of LIF improved PR survival and retinal function by activating STAT3 in Müller glia and PR (103). Stressed PRs secrete signal molecules such as Edn2 and H2O2 that facilitate LIF induction in Müller glia; Edn2 triggers LIF transduction by binding to endothelin receptor B (Ednrb) localized to Müller glia; and H2O2 increases LIF transcript levels by stabilizing LIF mRNA via ILF3 (interleukin enhancer binding factor 3) (98, 218–220). LIF deficiency or Ednrb antagonism diminishes JAK/STAT activation and the amount of reactive Müller glia, resulting in accelerated degeneration; in contrast, LIF supplementation or Ednrb agonism improves PR survival in degenerating retina (218, 221). Deletion of gp130 in either Müller glia or rod PRs severely dampened the activation of CNTF-triggered signaling as well as PR rescue (102), and when Müller glia were ablated, LIF no longer provided protection (222). However, other research suggests that gp130 deficiency in Müller glia decreases STAT3 phosphorylation but does not weaken the neuroprotection of exogenous LIF (223) because gp130 activation in PR presumably mediates a cell-autonomous protective mechanism with a general protective role independent of pathological stimulus (223, 224). Modulation of JAK/STAT signaling results in contrary immunomodulatory effects in different retinal components. On the one hand, inhibition of JAK2/STAT3 in microglia contributes to inflammation mitigation. On the other hand, LIF-induced STAT3 signaling in Müller glia favors neuroprotection, which seems to be an endogenous protective mechanism. We speculate that this paradoxical outcome involves crosstalk between retinal microglia and Müller glia, which is not yet fully understood. Phosphorylated JAK also activates PI3K, so there may be synergy between JAK/STAT signaling and other pathways.
Epigenetic modifications, which include DNA methylation, histone modification, and noncoding RNAs, refer to changes in gene expression patterns without altering the genomic DNA sequence (225). Epigenetic modifications are implicated in aspects of individual growth and disease development, including gene expression, cell proliferation and differentiation, misfolded protein response, and cytoskeletal dynamics (226). Although the concept of curing diseases through epigenetic regulation is relatively new, it has demonstrated considerable therapeutic potential in research on cancer, autoimmune diseases, endocrine diseases, congenital disease and many others (227, 228). Epigenetic changes contribute to the development of RP, and remarkable progress has been made in the treatment of RP with epigenetic modification therapies.
Histone acetylation and methylation are the two most well-studied types of histone modification, with acetylation typically resulting in increased gene expression and methylation being related to either increased or decreased gene transcription. Histone acetylation is regulated by histone acetyltransferases and histone deacetylases (HDACs), while histone methylation is regulated by lysine methyltransferases and arginine methyltransferases and histone demethylation by histone demethylases. Enzymes that add or remove epigenetic marks on histones are known as “writers” and “erasers.” In addition, there are “readers” containing bromodomains, chromodomains, or Tudor domains that are able to decipher histone codes (229). RP retinas exhibit excessive HDAC activity (104, 105, 230), and HDAC inhibition delays retinal degeneration in RP animal models (rd1 and rd10 mice and zebrafish) (104–107). In rd10 mice, the HDAC inhibitor romidepsin prevented rod degeneration and enhanced retinal function. Two molecular mechanisms contribute to this neuroprotective effect. First, by acting on histone targets in PRs, increasing chromatin accessibility and upregulating neuroprotective genes, and second, by acting on nonhistone targets in microglia and resident and invading immune cells, it suppresses inflammatory gene transcription and inflammation (108). Microglial activity is related to histone methylation levels. LPS-activated microglia increase HDAC expression, which is accompanied by an increase in inflammatory gene expression (231). HDAC inhibition or knockdown promotes a protective microglial phenotype and reduces neuroinflammation (232–234). Valproic acid is an HDAC inhibitor that reduces PR degeneration in rd1 and P23H RP models (109, 110). Valproic acid increases the expression of STAT1 by inhibiting HDAC3 expression; subsequently, acetylated STAT1 forms a complex with nuclear NF-κB p65, preventing NF-κB p65 DNA-binding activity (235). Moreover, suppression of the “read” (bind) behavior to histone acetylation marks of bromodomain and extraterminal domain proteins by JQ1 ameliorated PR degeneration and maintained electroretinographic function in rd10 mice. This protection seems to be partially mediated by the inhibition of retinal microglial proliferation, migration, and cytokine production (111). Several studies, including our previous report, have reported altered histone methylation in RP retinas (112, 236, 237). Lysine demethylase 1 inhibition attenuated PR degeneration in rd10 mice, in part by inhibiting microglial-related inflammation (108). DZNep (3-deazaneplanocin A) specifically inhibits Ezh2 (H3K27 trimethyltransferase) and mediates neuroprotective effects in rd1 mice by inhibiting H3K27me3 deposition (112). Ezh2 reportedly mediates TLR-induced inflammatory gene expression (238) and activation of multiple types of inflammasomes in microglia (239), hence promoting microglial-related pathologies.
MicroRNAs (miRNAs) are small noncoding RNAs that modify gene expression post-transcriptionally by targeting messenger RNAs, long noncoding RNAs, and pseudogenes and circular RNAs. MiRNAs can be packed into exosomes or microvesicles to perform long-distance cell-to-cell communication. MiRNAs play a critical role in gene expression modulation and are therefore interesting candidates for the development of biomarkers and therapeutic targets (240). Throughout development, miRNAs are required for retinal neuron differentiation (241, 242). Dysregulated miRNAs were found in the retinas of mouse and canine models of RP (243, 244), indicating the involvement of miRNAs in the etiology of RP. MiRNAs regulate microglial phenotypes, as evidenced by various studies on retinal and neurodegenerative disorders (245, 246). Inhibition of miR-6937-5p preserved the outer nuclear layer thickness and promoted the ERG wave response in rd10 mice (113), and AAV-miR-204 attenuated retinal degeneration in two different mouse models. By downregulating microglial activation and PR mortality, miR-204 alters the expression profiles of transgenic retinas toward those of healthy retinas (114). In addition, miR-223 is required for the regulation of microglial inflammation and the maintenance of normal retinal function (247).
DNA methylation refers to the addition of a methyl group to the 5′-carbon of a cytosine (C) ring, resulting in the formation of 5-methylcytosine (5mC), which mainly occurs in the promoter regions. Typically, methylation modifications result in gene repression, and global genomic hypermethylation relates to heterochromatin formation and inhibits transcription (248). Aberrant regulation of DNA methylation results in PR degeneration and neuronal loss in the retina. In the absence of DNA methyltransferase 1, the initiation of PR differentiation is severely hindered (249). In RP retinas, binding sites of several important transcription factors for retinal physiology were hypermethylated (250). The role of DNA methylation in the development of retinitis pigmentosa has been reviewed in detail elsewhere (251) and will not be repeated here. We argue that trained immunity regulates the microglial phenotype in RP by plasticizing microglial reactivity via epigenetic modification. Trained immunity, also known as innate immune memory, refers to the phenomenon in which innate immunity modifies its function after an initial insult and reacts more vigorously to subsequent stimuli. Epigenic reprogramming determines the immune phenotype of immune cells and leads to long-lasting functional alterations (252, 253) (Figure 2). Using macrophages as an illustration, in the resting state, the promoter regions of inflammatory genes are enriched with repressive epigenetic marks, called epigenetic barriers, to prevent activation in the absence of stimuli. Upon stimulus, repressive epigenetic marks are removed, and activating epigenetic marks are introduced to the promoters and enhancers of specific genes in an attempt to encourage inflammatory molecule synthesis and phagocytosis to eliminate the insult. After stimulus elimination, activating epigenetic marks are partly retained (254). The innate immune system may become overly trained in chronic inflammatory diseases as a result of such mechanisms, resulting in pathological tissue damage. In the context of neurodegenerative disorders of the CNS, the relationship between trained immunity and microglial phenotype has been discussed (255, 256). Low-dose LPS intraperitoneally administered to mice induced long-lasting innate immune memory in brain microglia and exacerbated Alzheimer’s disease pathology. Activated microglia are enriched with the epigenetic marks H3K4me1 and H3K27ac, which define active enhancers (256). In RP model P23H rats, intraperitoneal injection of low-dose LPS increased microglial activation and the number of infiltrating microglia, as well as elevated the expression levels of several inflammation-related genes (257). In addition to the activation of retinal microglia, elevated levels of serum cytokines show the activation of peripheral immune cells in RP (22, 23). Recent work by Su et al. revealed that monocytes from patients with autosomal recessive RP exhibit a trained-like phenotype. Upon stimulation, these monocytes produce more TNF-α, IL-6, and IL-1β and upregulate inflammatory pathways such as NF-κB (258). Current evidence supports a role for trained immunity in RP pathogenesis by epigenetic reprogramming of microglia and peripheral macrophages to modulate the immune phenotype and trigger an active immune response, although many details remain to be confirmed.
The gut microbiome, which resides in the intestinal tract and performs nutrition metabolism, has recently been found to influence the maturation of the immune system. Components and metabolites of microbial cells engage in the modulation of immune recognition and immune tolerance through innate immune receptors on intestinal epithelial cells and influence the function of innate myeloid cells and lymphoid cells through diverse mechanisms (Figure 3). In addition, the microbiota’s make-up and function are subject to the innate immune system. Therefore, gut dysbiosis may induce immune system dysregulation and trigger disease emergence (259). The gut microbiome has been linked to retinal degenerative disorders such as age-related macular degeneration (260) and diabetic retinopathy (261). Using the rd10 RP mouse model, Kutsyr O. et al. (262) related alterations in the composition profiles of the gut microbiome to RP. Compared to healthy mice, the gut microbiome of rd10 mice had reduced ASV richness and α diversity. Rd10 mice, in particular, feature a high proportion of B. caecimuris, a species that is uncommon in healthy gut mice, but lack four species (Rikenella spp., Muribaculaceace spp., Prevotellaceae UCG-001 spp., and Bacilli spp.) that are common in the healthy gut microbiome (262). The gut microbiome is susceptible to dietary influences. Further research by the same group demonstrated that a short-term high-fat diet significantly modifies the gut flora, enhances retinal oxidative stress and inflammation, and ultimately accelerates the degeneration of the rd10 retina (263). Thus, dysbiosis in the gut contributes to retinal inflammation and constitutes the pathogenesis of RP. By exchanging the intestinal microbiota (Fecal microbiota transplant, FMT) of young and aged mice, emerging evidence by Parker et al. (264) suggests that the gut microbiome is a modifier of retinal inflammation. Compared to young mice, aged mice exhibit increased systemic and tissue inflammation, as evidenced by elevated serum proinflammatory cytokines (TNFα, IL-6), microglial overactivation in the brain, and C3 accumulation at the RPE/Britch’s membrane interface. Transferring aged donor microbiota to young mice disrupts the intestinal epithelial barrier and triggers inflammation in the retina and brain, whereas transfer of aged mice with young donor microbiota could reverse age-related inflammation (264). The evidence above supports the “diet-gut microbiome-retina axis” hypothesis in the pathogenesis of RP. Despite the fact that this work is still in its early stages, the gut microbiota is a promising therapeutic target for RP.
Herbal compounds, or phytochemicals derived from plants, possess a wide range of biological activities and have been explored for the treatment of RP, demonstrating anti-inflammatory properties in RP investigations. Curcumin is a polyphenolic compound produced from the spice turmeric. Curcumin provided morphological and functional protection in rd1 mice, P23H rats, and an MNU-induced RP model (115, 116, 265, 266). A single vitreous injection of curcumin reduced PR loss in rd1 mice by inhibiting microglial activation and modulating the expression of CCL2, TIMP-1 and VCAM-1 (115). Lyceum barbarum polysaccharides and zeaxanthin dipalmitate are two main bioactive agents extracted from wolfberry. Lyceum barbarum polysaccharides protects against retinal degeneration by modifying inflammation and apoptosis through the inhibition of NF-κB and HIF-1α expression (117, 118). zeaxanthin dipalmitate acts through several pathways, including STAT3, CCL2 and MAPK, in parallel to inhibiting inflammation in the rd10 retina (119). Saffron, widely used in traditional Chinese medicine for its anti-inflammatory and antioxidant properties, protects PRs exposed to environmental ATP by blocking P2X7R signaling (120). In P23H rats, saffron administration increased PR survival and functional retention while decreasing vascular disruption (121). Resveratrol (3,40,5-trihydroxystilbene) is found in chocolate, fruits, and vegetables. Resveratrol treatment inhibited microglia-mediated death of 661W cells via downregulation of microglial migratory, phagocytic, and proinflammatory cytokine production (122). Subretinal injection of JC19 (3,4’-diglucosyl resveratrol), a resveratrol prodrug, reduced PR loss and improved functional performance in ERG tests of the rd10 retina. The author speculates that sirtuin1 activation is the underlying mechanism (123).
A large body of research conducted on the inflammatory processes during RP tries to discover common mechanisms that target multiple RP genotypes and develop appropriate therapeutic options. However, after reviewing the existing literature, we discovered that no single treatment is appropriate for all types of RP, and the application of valproic acid is a prime example, with treatment effects significantly varying between models with different genetic backgrounds and even exhibiting detrimental effects. This raises the prospect that a link between genetics and RP inflammation needs more investigation. To date, genetic mutations remain the only identified risk factor for RP. Different permutations of inheritance pattern, genotype, and the number of mutations lead to variations in the phenotype and pathological progression of RP. Similarly, we anticipate that the multiple phenotypes of inflammatory activation in RP are closely related to the genetic background. Nevertheless, the relationship between genetic background and inflammation is currently unclear due to the lack of corresponding evidence. Both RP patients and animal models have a more susceptible immune system and are prone to developing inflammation. This abnormal immune system may depend heavily on the genetic background. The gut microbiota play a critical role in the maturation of the innate immune system after birth, and trained immunity is implicated in this process; however, the influence of genetic background on the maturation of the immune system has not been investigated. Therefore, long-term clinical observation and family tracing of the RP population are necessary. What needs to be documented should include, but is not limited to, macroscopic clinical manifestations, structural and functional measurements, and monitoring of local and peripheral inflammation levels. And appropriate follow-up criteria need to be established to ensure consistency of measurements and to obtain usable information. Inflammation is an important feature of RP, and the present review highlights the role of immunomodulation in RP treatment. There has been significant interest in modulating the inflammatory response as a strategy to treat RP, and an increasing number of studies have proven the effectiveness of immunomodulation in ameliorating and perhaps reversing retinal degeneration. Therapeutic strategies based on immunomodulation are a potential treatment for RP, and deepening the understanding of immune modulation is helpful in establishing suitable therapies. As with immunotherapies already carried out, artificial regulation of immunity will bring inevitable side effects. It is challenging to regulate immunity accurately and to enhance the beneficial effects and minimize the harmful ones concurrently. Many of these specific mechanisms need to be further studied, especially the interactions between these pathways.
LZ wrote the manuscript and painted the figure. NY and CH reviewed and modified the article. All authors contributed to the article and approved the submitted version.
This work was supported by grants from the Science and Technology Department of Sichuan Province (2020YFS0205).
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647060 | Yongjuan Li,Boyong Liao,Yi Wang,Huihua Luo,Shimin Wang,Caiqin Li,Wenpei Song,Kunchang Zhang,Boqun Yang,Shaoqiang Lu,Bipei Zhang,Yongquan Li | Transcriptome and metabolome analyses provide insights into the relevance of pericarp thickness variations in Camellia drupifera and Camellia oleifera | 27-10-2022 | Camellia drupifera,Camellia oleifera,pericarp thickness,lignin,transcriptome,metabolome | Camellia fruit is a woody edible oil source with a recalcitrant pericarp, which increases processing costs. However, the relevance of pericarp thickness variations in Camellia species remains unclear. Therefore, this study aimed to identify pericarp differences at the metabolic and transcription levels between thick-pericarp Camellia drupifera BG and thin-pericarp Camellia oleifera SG. Forty differentially accumulated metabolites were screened through non-targeted UHPLC-Q-TOF MS-based metabolite profiling. S-lignin was prominently upregulated in BG compared with SG, contributing to the thick pericarp of BG. KEGG enrichment and coexpression network analysis showed 29 differentially expressed genes associated with the lignin biosynthetic pathway, including 21 genes encoding catalysts and 8 encoding transcription factors. Nine upregulated genes encoding catalysts potentially led to S-lignin accumulation in BG pericarp, and transcription factors NAC and MYB were possibly involved in major transcriptional regulatory mechanisms. Conventional growth-related factors WRKYs and AP2/ERFs were positively associated while pathogenesis-related proteins MLP328 and NCS2 were negatively associated with S-lignin content. Thus, Camellia balances growth and defense possibly by altering lignin biosynthesis. The results of this study may guide the genetic modifications of C. drupifera to optimize its growth–defense balance and improve seed accessibility. | Transcriptome and metabolome analyses provide insights into the relevance of pericarp thickness variations in Camellia drupifera and Camellia oleifera
Camellia fruit is a woody edible oil source with a recalcitrant pericarp, which increases processing costs. However, the relevance of pericarp thickness variations in Camellia species remains unclear. Therefore, this study aimed to identify pericarp differences at the metabolic and transcription levels between thick-pericarp Camellia drupifera BG and thin-pericarp Camellia oleifera SG. Forty differentially accumulated metabolites were screened through non-targeted UHPLC-Q-TOF MS-based metabolite profiling. S-lignin was prominently upregulated in BG compared with SG, contributing to the thick pericarp of BG. KEGG enrichment and coexpression network analysis showed 29 differentially expressed genes associated with the lignin biosynthetic pathway, including 21 genes encoding catalysts and 8 encoding transcription factors. Nine upregulated genes encoding catalysts potentially led to S-lignin accumulation in BG pericarp, and transcription factors NAC and MYB were possibly involved in major transcriptional regulatory mechanisms. Conventional growth-related factors WRKYs and AP2/ERFs were positively associated while pathogenesis-related proteins MLP328 and NCS2 were negatively associated with S-lignin content. Thus, Camellia balances growth and defense possibly by altering lignin biosynthesis. The results of this study may guide the genetic modifications of C. drupifera to optimize its growth–defense balance and improve seed accessibility.
Camellia drupifera, a newly identified Camellia species serving as a woody edible oil crop similar to oil palm, olive, and coconut (Long et al., 2008; Wang et al., 2014; Chen et al., 2015), is grown specifically in South China and genetically proximal to Camellia oleifera (Flora of China, 1994; Qin et al., 2018). Oil extracted from Camellia seeds is rich in monounsaturated fatty acids and other bioactive metabolites with anticancer, antioxidant, and immunity-enhancing effects, thereby becoming a healthy high-grade edible oil in the global market these days (Suealek et al., 2021; Liu et al., 2022). Increased attention has been drawn to cost-efficient oil production from seeds that are embedded in dry fruits or capsules. The capsule of Camellia is anatomically composed of seeds, pericarp, and carpel bundles. Dry dehiscent capsules are split via three to five valves, with each section holding one to four seeds when the fruit is ripe (Patrick, 1997; Ming, 2007; Orel and Wilson, 2012). Oil-rich seeds are under firm protection of pericarps against external abiotic and biotic stresses before maturation; however, the recalcitrance of woody pericarps (a matrix of lignocellulosic materials) increases processing costs for seed accessibility (Li et al., 2016). Therefore, pericarp thickness has been considered an economic structural trait of Camellia species, where thin ones are favored over thick ones. Tan et al. (2020) identified the constituents of woody pericarps in C. oleifera Abel to be 15.8% cellulose, 23.6% hemicellulose, 8.8% lignin, and others, which include polymers deposited in secondary cell walls. Lignin polymers cross-link with cellulose microfibrils and hemicellulose molecules via side chains, forming a rigid cell skeleton and rendering the pericarp recalcitrant (Reddy et al., 2005; Chen and Dixon, 2007; Henry, 2010). Different amounts and/or compositions of lignocellulosic constituents produce distinct pericarp thicknesses in Camellia species. However, the relevance of pericarp thickness variations in Camellia species and the underlying regulatory mechanisms remain unclear, complicating the genetic modification of this trait. Genes involved in the biosynthesis of some components, particularly cellulose and lignin, have been characterized in other plants. The pericarp of C. drupifera capsules is notably thicker than that of C. oleifera. Both species are endemic to South China, providing ideal materials for investigating the relevance of pericarp thickness variations in Camellia species. Apart from pericarp thickness, some coupled properties related to growth and defense have also aroused wide concern. For instance, thick pericarps are usually accompanied by tall trunks and large capsules but weak defense, whereas trees with thin pericarps always exhibit strong defense but dwarf trunks and small capsules (Yang et al., 2015; Hong et al., 2016). The present study aimed to compare the metabolic and transcriptional profiles of capsules from thick-pericarp C. drupifera BG (hereafter BG) and thin-pericarp C. oleifera SG (hereafter SG) and elucidate the mechanism by which pericarp thickness influences growth–defense tradeoffs in Camellia. In the current study, we found that syringyl lignin (S-lignin) was significantly upregulated in BG compared to SG, resulting in a thicker peel in BG. These results provide insights into the molecular basis of pericarp thickening in Camellia species. This study may guide the genetic modifications of C. drupifera to optimize its growth–defense balance and facilitate seed accessibility.
Thick-pericarp C. drupifera BG and thin-pericarp C. oleifera SG were planted in the Boluo Forest Farm and the Xiaokeng Forest Farm Guangdong Province, China, under the same growth conditions. Fruit samples from individual trees were collected randomly from different branches in November 2020. Fresh capsules were dissected using a blade on ice bed, and their pericarps were separated manually while wearing sterile gloves, frozen immediately in liquid nitrogen, and then stored at -80°C for RNA extraction and metabolite analysis. Three biological replicates of each sample were used for RNA sequencing, and eight biological replicates were used for metabolic profiling. In the present study, the pericarp samples of the BG and SG capsules were abbreviated as BG_1, BG_2, BG_3… SG_1, SG_2, SG_3…
The transverse, longitudinal, and horizontal diameters and pericarp thickness of 25 capsules were measured with a vernier caliper, and the fresh weight of seeds, fresh weight of pericarps, and weight of a single capsule were determined. The pulp at the transverse diameter was selected for the determination of pericarp thickness. A comparative analysis of the capsule phenotypic characteristics of BG and SG was performed using Excel software.
The cellulose and hemicellulose contents of 0.5 g samples were prepared and quantified as previously described by Yan et al. (Yan, 2020). Lignin was assayed by derivatization with acetyl-bromide-glacial acetic acid (Barnes and Anderson, 2017). The pericarp was ground and filtered through a 60-mesh screen, and 5 mg of pericarp powder was added to 10 mL of 10% acetyl bromide–glacial acetic acid solution. and then, 500 μL of 70% perchloric acid was added and heated at 50°C for 30 min. After cooling, the reaction was terminated by mixing 10 mL of 2 mol/L NaOH and 10 mL of glacial acetic acid. Subsequently, the samples were centrifuged again at 8000 × r for 5 min. Absorbance was subsequently obtained at 280 nm using glacial acetic acid as a control.
For lignin characterization, the pericarps were stained with toluidine blue O (TBO) reagent (Soleibo Technology Co., Ltd., Beijing, China) as described previously (Jia et al., 2015). Pericarp specimens were cut from the middle part of the capsules and then soaked overnight in 70% formaldehyde alcohol acetic acid fixative solution. Paraffin sections were immediately stained with 0.05% (w/v) TBO for 10 min. The specimens were washed, dehydrated with an ethanol series of 75%–95%, and then embedded in neutral resin. The samples were sectioned using an ultrathin semiautomatic microtome (Lerca-RM2235, Germany) to prepare paraffin sections in accordance with the manufacturer’s instructions. Micrographs were taken under a Leica microscope (DM2000 LED, Germany).
Freeze-dried samples were crushed using the liquid-nitrogen grinding method. Powdered tissue (100 mg) was dissolved in 1.0 mL of cold methanol/acetyl cyanide (50:50 v/v) and blended twice by low-temperature ultrasonic treatment for 30 min. Extraction was stable at -20°C for 60 min, and then the homogenate was centrifuged at 14,000 × g for 20 min at 4°C. The supernatant was used for Liquid Chromatography-Mass spectrometry. Metabolites were analyzed using an ultra-performance liquid chromatography system (UHPLC, Agilent 1290 Infinity LC system, USA) equipped with a HILIC column (1.7 μm, 2.1 mm× 100 mm column). The column temperature was maintained at 25°C, the flow rate was 0.3 mL/min, and the injection volume was 2 μL. Water with 0.04% ammonium acetate (v/v) and 0.04% ammonia and acetonitrile were used as the compositions of mobile phases A and B, respectively. Gradient elution was performed as follows: 0–1 min, 85% B (v:v); 1–12 min, 85%–65% (v:v); and 12–12.1 min, 65%–40% B (v:v), 12.1–15 min, 40%–85%, 15.1–20 min, 85% B, with automatic injection at 4°C during the whole analysis. Primary and secondary spectra of the samples were collected using an AB Triple TOF 6600 mass spectrometer (AB Sciex, Concord, Canada). The ESI source conditions after HILIC chromatographic separation were as follows: Ion Source Gas1, 60 psi; Ion Source Gas2, 60 psi; curtain gas, 60 psi; source temperature, 600°C; Ion Sapary Voltage Floating ±5500 V (positive and negative modes); TOF MS scan m/z range, 60–1000 Da; product ion scan m/z range, 25–1000 Da, TOF MS scan accumulation time, 0.20 s/spectra; product ion scan accumulation time, 0.05 s/spectra; and declustering potential (DP), ± 60 V (positive and negative modes). Raw MS data (.wiff scan files) were converted to mzML files using ProteoWizard (Chambers et al., 2012). The peak alignment, retention time alignment, and peak area were processed using XCMS software. The following parameters were used for peak picking: centWave, m/z = 25 ppm; peak width, c (10, 60); and prefilter, c (10, 100). The following parameters were used for peak grouping: bw, 5; mzwid, 0.025; and minfrac, 0.5. Metabolite identification was based on an in-house database at Shanghai Applied Protein Technology Co. Ltd. (Shanghai, China) established with authentic standards. After normalization to total peak intensity, the processed data were uploaded into SIMCA-P14.1 (Umetrics, Umea, Sweden) for mode identification, where they were subjected to multivariate data analyses, including principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA). The variable importance in projection (VIP) scores of each variable within the OPLS-DA model were calculated to indicate their contribution to their classification.
Total RNA was extracted from the BG and SG capsules by using TRIzol Reagent (Magen, Guangzhou, China) and purified using the AMPure XP system (Beckman, USA). Paired-end libraries were prepared using an ABclonal mRNA-seq Lib Prep Kit (Abclonal, China) following the manufacturer’s instructions. Sequencing was performed using an Illumina NovaSeq 6000 instrument (Shanghai Applied Protein Technology Co. Ltd., Shanghai, China). Raw data in fastq format were processed using Perl scripts. Clean reads were obtained by removing adapter sequences, low-quality reads (number of lines with a string quality value less than or equal to 25 accounts for more than 60% of the entire reading), and reads with N ratio (base information cannot be determined) greater than 5%. High-quality reads were assembled into contigs, transcripts, and unigenes by using Trinity software (http://trinityrnaseq.sourceforge.net/). FeatureCounts (http://subread.sourceforge.net/) was used to count the number of reads mapped to each gene. Then, the FPKM of each gene was calculated based on the length of the gene and the read count mapped to this gene. Differential expression analysis was performed using DESeq2 (http://bioconductor.org/packages/release/bioc/html/DESeq2.html) (Love et al., 2014). For functional annotation and classification, the assembled transcriptome sequences were compared to obtain annotation information in each of the five following databases: NR (http://ftp.ncbi.nlm.nih.gov/blast/db/) (Deng et al., 2006), Swiss-Prot (http://web.expasy.org/docs/swiss-prot.guideline) (Rolf et al., 2004), Pfam (http://pfam.xfam.org/) (Finn et al., 2014), gene ontology (GO, http://www.geneontology.org) (Ashburner et al., 2000), and Kyoto Encyclopedia of Genes and Genomes (KEGG, http://www.genome.jp/kegg/) (Minoru et al., 2016). GO and KEGG enrichment analyses were conducted using the clusterProfiler R software package to explain the functional enrichment of DEGs and clarify the differences between samples at the gene function level. The GO or KEGG functions were significantly enriched when P < 0.05.
The reads count from RNA-seq reads and the sinapyl alcohol value from metabolome data were calculated correlation and p-value using the R platform (version 4.0.5). Then, lignin-related metabolic compounds and catalytic enzymes in Camellia and other plants were investigated from NR, Pfam, Swissprot, GO and KEGG databases and extracted from the calculated results. Co-expression network patterns were visualized by cytoscape (version 3.9.1, Shannon et al., 2003).
Single-stranded cDNAs were synthesized from the RNAs using the PrimeScript™ RT reagent Kit, and quantitative real-time PCR was performed using ViiA7 (Thermo Fisher Scientific, USA) and Hieff™ qPCR SYBR Green Master Mix (Yisheng Biotechnology, Shanghai, China). Primers were designed by Bioruqi (Guangzhou, China) and synthesized by GENEWIZ (Suzhou, China). The glyceraldehyde-3-phosphate dehydrogenase gene (GAPDH) was used as an internal reference, and the relative expression was calculated using the 2ΔCt method. The standard errors of the means among the replicates were calculated. All quantitative real-time polymerase chain reaction (qRT-PCR) analyses were performed in three biological replications, respectively. The expression patterns of eight transcripts were monitored, and detailed information about the unigene IDs, fold change (FC), and primer pairs designed in this study are presented in Supplementary Table S1 .
Raw transcriptome data from BG and SG were uploaded to the SRA database (https://www.ncbi.nlm.nih.gov/sra). Other raw transcriptome data of five sibling Camellia species, including C. japonica (SRR17275277), C. chekiangoleosa (SRR15420647), C. petelotii (SRR17460028), C. azalea (SRR7120561), and C. oleifera (SRR17365493), were downloaded from the SRA database. Paired-end sequencing datasets were first trimmed with Trimmomatic (0.39) (http://www.usadellab.org/cms/?page=trimmomatic) using default settings, and then quality checks were performed using fastQC (0.11.9) (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) to confirm the removal of adapters and low-quality regions. The obtained sequences were then assembled using Trinity (2.9.1). Protein sequences were deduced using ORFfinder (https://ftp.ncbi.nlm.nih.gov/genomes/TOOLS/ORFfinder). All parameters were set to their default values. Single-gene orthologs were identified using OrthoFinder to cluster the protein sequences from the seven species (Emms and Kelly, 2019). Multiple sequence alignments were performed based on the amino acid sequences using the alignment tool MUSCLE with default parameter settings. Maximum-likelihood phylogenetic trees were constructed using MEGA7 software with the JTT model (Sudhir et al., 2016). In the phylogenetic tree, bootstrap supporting values below 50 were generally regarded as unreliable and are not shown. The 334 single-copy nuclear genes were conserved in eggNOG (http://eggnog5.embl.de/#/app/home) (Huerta-Cepas et al., 2019). The pictures of other Camellia species are from the plant plus of China (http://www.iplant.cn).
We quantitatively characterized and compared the pericarp thicknesses of BG and SG capsules at mature developmental stages. As shown in Figures 1A, B , the BG capsules had a thicker pericarp than the SG capsules. Further quantification of capsule size in terms of cross and longitudinal diameters, pericarp thickness, and pericarp proportion supported the different capsule phenotypes observed between the BG and SG samples ( Figures 1C, D ). Moreover, lignin content, which contributes to mechanical support, was approximately 30% higher in the BG capsules than in the SG capsules. Although cellulose and hemicellulose were the main components of secondary cell walls, their contents did not significantly differ between the BG and SG capsules ( Figure 1E ). These results indicate that the higher lignin deposition in the capsule cell walls of BG than those of SG is solely responsible for the thicker pericarps of BG than SG.
We analyzed the metabolite profiles of the BG and SG capsules and determined their correlation with pericarp thickness. Nontargeted UHPLC-Q-TOF MS-based metabolite profiling in positive and negative ion modes revealed 318 metabolites. These metabolites were classified into eight groups and included eight major compounds in the monolignol biosynthetic pathway ( Figure 2B and Supplementary Table S2 ). The first component of the PCA results (46%) predominantly reflected the difference between SG and BG, and the second component (10.2%) primarily indicated within-group differences, which separated the BG and SG samples in terms of metabolite composition and content ( Figure 2A ). The differentially accumulated metabolites (DAMs) were represented based on variable importance in the projection score (VIP > 1.0), fold change (FC > 2.0 or FC < 0.5), and p-value (p < 0.05. To eliminate the effects of quantity on pattern recognition, we applied Z-score transformation of the peak areas for each metabolite and subsequently performed hierarchical cluster analysis. As shown in Figures 2B–D , sinapyl alcohol (S-unit, one generic primary monolignol) and organic oxygen compounds were the most upregulated metabolites while organic acids and derivatives were the most downregulated metabolites in the BG capsules. Despite the fact that coniferyl alcohol (G-unit) and coumarate conjugate (H-unit) were also viewed as essential monomers of lignin, the former (G-unit) showed no significant difference in amount between the BG and SG capsules, and the latter (H-unit) was undetectable in both species ( Supplementary Table S2 ). Sugar units, as typical monomers of cellulose and hemicellulose, were at similar levels in the two species, suggesting that the accumulation of S-lignin was the key factor contributing to the distinct pericarp thickness of the two species.
The transcriptome profiles of BG and SG were obtained and compared to investigate the transcriptional regulation of DAMs in pericarps from the BG and SG samples. Approximately 245.07 Gb of raw data was generated, and the statistics of the sequencing libraries are summarized in Table 1 . Unique-mapped reads were used to calculate the expression levels in transcripts per million (Wagner et al., 2012; Vera Alvarez et al., 2018). The resulting sets yielded 2.5 × 108 clean reads, with over 60% mapped to the assembly transcripts from RNA-seq data in non-model organisms ( Table 1 ). A total of 337,768 protein-coding genes were predicted in NR, Swissprot, PFAM, GO, and KO, of which 8974 DEGs; 4322 upregulated and 4652 downregulated) were identified with the criteria of a false discovery rate (FDR) < 0.05 and a |log2FC | > 1 ( Figure 3A and Supplementary Table S3 ). GO functional enrichment analyses of DEGs were assigned to biological processes (72.2%), molecular functions (87.0%), and cellular components (64.8%). Among the biological processes, metabolic and macromolecular biosynthetic processes were highlighted. Meanwhile, most of the DEGs were enriched in “structural molecule activity” and “structural constituent of ribosome” molecular function terms toward structural molecular activity, which is consistent with the DAMs profile ( Figure 3B and Supplementary Table S4 ). qRT-PCR results validated the RNA-seq results obtained using eight monolignol biosynthesis-related DEGs. The expression patterns of these genes obtained using qRT-PCR were consistent with those determined by RNA-seq ( Figure 3C ). For the nuclear Camellia phylogeny, we used transcriptomic sequences of BG and SG and combined them with public transcriptome datasets from six other Camellia species to identify 334 conserved single-copy nuclear genes ( Supplementary Table S5 ). The smallest gene set was then subjected to phylogenetic analysis using the maximum-likelihood method, which highly supported a divergence of BG and SG after the most recent common ancestor of C. oleifera, suggesting the feasibility of frequent interspecific hybridization and genetic introgression between the two species during the evolution and domestication of Camellia ( Figure 3D ).
To provide an overview of the DAMs and DEGs involved in the lignin biosynthetic pathway, we mapped the metabolic compounds and catalytic enzymes onto the pathway based on the KEGG database. A moderate proportion of the genes and compounds involved in the lignin biosynthetic pathway, including precursors, intermediates, end products, and sequential enzymes, were distinct between the SG and BG capsules ( Figure 4A ). Based on the KEGG database, we identified 11 catalytic enzymes in Camellia that are potentially involved in the biosynthetic pathway to the major monolignol precursors of lignin. The expression levels of genes encoding p-coumarate 3-hydroxylase (C3’H), caffeoyl shikimate esterase (CSE), and caffeic acid O-methyltransferase (COMT) were increased in BG, leading to a predominance of S-units (81.4%), accompanied by low levels of G-units (18.6%) despite reduced expression levels of cinnamyl alcohol dehydrogenase (CAD), cinnamate 4-hydroxylase (C4H), and L-phenylalanine ammonia-lyase (PAL). The oxidative polymerization of monolignols is catalyzed by peroxidases (PODs, using hydrogen peroxide) and laccases (LACs, using molecular oxygen) (Liu et al., 2018; Dixon and Barros, 2019). We also identified four POD genes and one LAC gene in Camellia that were homologous to AtPRX and AtLAC15 in Arabidopsis, which might enhance monolignol bulk polymerization in BG through their upregulated expression ( Figure 4A and Supplementary Table S6 ). To further explore the transcription factors (TFs) regulating lignin metabolism in Camellia, we performed a coexpression network analysis to direct differentially expressed TFs toward genes encoding S-lignin closely related enzymes (i.e., COMT1, POD2, CSE, POD1, C3’H, POD3, and POD4), and then narrowed down eight hub TFs into a highly correlated key module (|correlation| > 0.9) ( Figure 4B ). Among these TFs, wax inducer1/SHINE1 (WIN1/SHN) and ethylene response factor 38 (ERF38), which belong to the AP2/ERF family, were positively correlated with lignin metabolism in Camellia, as did MYB62, NAC29, SPATULA (SPT), and WRKY transcription factor 44 (WRKY44), whereas MLP328 and NCS2, which belong to the pathogenesis-related (PR) protein family, were negatively correlated with all enzymes previously mentioned ( Figure 4B and Supplementary Table 6 ). Taken together, the results indicated that these TFs, based on their annotated orthologs in Arabidopsis, were hypothesized to be the master regulators (activators and/or repressors) of pericarp lignification in Camellia.
Camellia L. is a good oil feedstock because of the biomass accumulation ability of its seeds embedded in lignocellulosic pericarps; therefore, pericarps with thin and low-density lignocellulose are favorable because they make seeds accessible (Yan, 2020). The pericarp thickness variations between BG and SG primarily arise from their genetic distinction, providing insights into the growth–defense tradeoffs in Camellia against stresses during evolution and domestication.
Mature Camellia seeds are generally protected by a rigid pericarp (approximately 1.2–7.0 mm) against external hazards, and pericarp thickness is mainly determined by secondary cell wall components, such as cellulose, hemicellulose, lignin, and a small proportion of other components (Tan et al., 2020; Figure 1 ). The cellulose and hemicellulose contents of the BG and SG capsules were nearly the same, but the lignin content of the BG pericarp was higher than that of the SG pericarp, implying that lignin accumulation was responsible for the thick pericarp of Camellia ( Figure 1E and Supplementary Table S8 ). Lignin is the second most abundant biopolymer that polymerizes on the cell wall surface (Ralph et al., 2004). The three essential monolignols are p-hydroxyphenyl (H, derived from 4-coumaryl alcohol), guaiacyl (G, derived from coniferyl alcohol), and syringyl (S, derived from sinapyl alcohol) units (Boerjan et al., 2003; Bonawitz and Chapple, 2010; Liu et al., 2018). Comparison of metabolic profiles showed that S-unit was the only differentially accumulated monolignol among the three main monolignols. Thus, the S/G ratio was higher in the BG capsules than in the SG capsules. This phenomenon was accompanied by the attenuation of flavonoid biosynthesis via the shikimate pathway and consequently increased synthesis of monolignols from shikimate and phenylalanine (Vogt, 2010; Figures 2B , 4 and Supplementary Table S2 ).
In this study, we demonstrated the biosynthetic pathway to the major monolignol precursors of lignin in Camellia: from phenylalanine via the phenylpropanoid pathway and subsequent monolignol polymerization via PODs and LACs ( Figure 4A ). The catalytic enzymes, as in most plants, comprise the deaminase PAL, which converts phenylalanine to cinnamic acid, and hydroxylases, methyl/acyl-transferases, reductases, and oxidases for polymerization (Liu et al., 2018; Dixon and Barros, 2019). We speculated that the high level of S-unit in the BG capsules was associated with activated shikimate shunt, which was catalyzed by the increased activities of C3’H and CSE, supplying an abundant shikimate pool despite the decreased activities of CADs (A. Wagner et al., 2007; Figure 4 ). Finally, oxidative polymerization ultimately determined the output of S-lignin, which was also likely boosted in the BG capsules for the increased activities of PODs and LACs ( Figure 4A ). The most widely studied TFs in the regulation of lignin biosynthesis are those belonging to the MYB and NAC families, including MYB58 and MYB63 in Arabidopsis, MYB31 in Musa, SND1 and NST1 in Arabidopsis, and NAC141 in Eucalyptus, which act as activators or repressors (Legay et al., 2007; Zhong et al., 2007; Sun et al., 2021). In the present study, MYB62 and NAC29 were screened through the coexpression analysis of the RNA-seq dataset for their potential role in regulating the lignin biosynthetic pathway in Camellia ( Figure 4B and Supplementary Table S7 ). Additionally, the expression of conventional growth-related factors WRKYs and AP2/ERFs, such as AtWRKY12 (Li et al., 2015), PtrWRKY19 (Yang et al., 2016), and OsSHN/WIN (Ambavaram et al., 2011), has been associated with lignin contents in some plants. Meanwhile, WRKY44 and WIN1 in Camellia, which are positively correlated with catalysts in the lignin biosynthetic pathway, might function as activators. SPT, which encodes a bHLH TF, was originally identified for its role in carpel and fruit development (Groszmann et al., 2008; Makkena and Lamb, 2013). Thus, it was presumed to divert the S-lignin biosynthetic metabolon specifically into Camellia pericarps in the present study. Intriguingly, two defense-induced genes, pathogenesis-related 2, were negatively correlated with these catalysts, which could explain the weak defense in BG.
Plant growth–defense tradeoffs involve resource reallocation to different biological processes, whereby plants optimize performance and fitness in a dynamic environment (Xie et al., 2018). Lignin biosynthesis is commonly believed to be an indivisible part of the complicated crosstalk between growth and defense because lignin provides mechanical strength, transports water and nutrients, and acts as a physical barrier to pathogen ingress; however, the relationship of lignin content with growth and defense remains largely elusive (Ha et al., 2021). Our previous study showed that S-lignin enrichment in BG pericarp co-occurs with increased growth rate and weak immunity, which differ from those in SG pericarp with low lignin content (our unpublished data). Similar findings have been reported in studies of Arabidopsis MYB46 and quinate/shikimate p-hydroxy cinnamoyl transferase (HCT) mutants, where reduced lignin content triggers stunt growth and enhances defense (Gallego-Giraldo et al., 2011). During the evolution and domestication of Camellia, frequent interspecific hybridization and genetic introgression between C. drupifera and C. oleifera would be conceivable because of their phylogenetically close relationship ( Figure 3D ). Thus, lignin content variation became prominent and was associated with the evolutionary adaptation of Camellia species. The large amount of energy invested in lignin accumulation can compensate for the costly expenditure of defense, producing a growth-bias phenotype. However, excessive lignin accumulation also inhibits growth. Thus, within a certain threshold, lignin content possibly acts as a “fulcrum” that balances the flux of resources, such as carbon and energy, into growth and defense, helping Camellia species regulate their response to environmental changes ( Figure 5 ). Whether or not the model we established about the lignin-dependent growth–defense tradeoffs in SG and BG can be extended to other Camellia species requires validation but could serve as a basis for further exploration.
The original contributions presented in the study are publicly available. This data can be found here: NCBI, PRJNA870661.
Data curation, investigation, visualization, writing–original and draft, YJL; Conceptualization, funding acquisition, project administration, writing–review and editing, BL; Formal analysis and writing–original draft, YW; Data curation, HL; Investigation, KZ, BY, and SL; Visualization, WS; Methodology, CL and WS; Conceptualization, supervision, writing-review, and editing, BZ; Funding acquisition, formal analysis and project administration, YQL. All authors have read, corrected, and approved the manuscript.
This research was funded by the Key-Area Research and Development Program of Guangdong Province, grant number 2020B020215003 and the Guangzhou Municipal Science and Technology Project, grant number 202201011754.
We thank Xiaokeng Forest Farm (Shaoguan City, Guangdong Province, China) and Boluo Forest Farm (Huizhou City, Guangdong Province, China) for providing plant materials. We thank Shanghai Applied Protein Technology Co. Ltd. For assisting us with the metabolic and transcriptome data analyses. We thank Bioruqi Technology Co., Ltd. (Guangzhou, China) for assisting us with qPCR experiments and Hunan Platte Network Technology Co., Ltd. For assisting us with toluidine blue reagent staining experiments.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647062 | Arnaud John Kombe Kombe,Fleury Augustin Nsole Biteghe,Zélia Nelly Ndoutoume,Tengchuan Jin | CD8+ T-cell immune escape by SARS-CoV-2 variants of concern | 27-10-2022 | SARS-CoV-2,cellular immunity,CD8 + T-cell epitope,cytotoxic T lymphocytes (CTL),variant of concern (VOC),protein mutation,HLA,immune escape | Despite the efficacy of antiviral drug repositioning, convalescent plasma (CP), and the currently available vaccines against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the worldwide coronavirus disease 2019 (COVID-19) pandemic is still challenging because of the ongoing emergence of certain new SARS-CoV-2 strains known as variants of concern (VOCs). Mutations occurring within the viral genome, characterized by these new emerging VOCs, confer on them the ability to efficiently resist and escape natural and vaccine-induced humoral and cellular immune responses. Consequently, these VOCs have enhanced infectivity, increasing their stable spread in a given population with an important fatality rate. While the humoral immune escape process is well documented, the evasion mechanisms of VOCs from cellular immunity are not well elaborated. In this review, we discussed how SARS-CoV-2 VOCs adapt inside host cells and escape anti-COVID-19 cellular immunity, focusing on the effect of specific SARS-CoV-2 mutations in hampering the activation of CD8+ T-cell immunity. | CD8+ T-cell immune escape by SARS-CoV-2 variants of concern
Despite the efficacy of antiviral drug repositioning, convalescent plasma (CP), and the currently available vaccines against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the worldwide coronavirus disease 2019 (COVID-19) pandemic is still challenging because of the ongoing emergence of certain new SARS-CoV-2 strains known as variants of concern (VOCs). Mutations occurring within the viral genome, characterized by these new emerging VOCs, confer on them the ability to efficiently resist and escape natural and vaccine-induced humoral and cellular immune responses. Consequently, these VOCs have enhanced infectivity, increasing their stable spread in a given population with an important fatality rate. While the humoral immune escape process is well documented, the evasion mechanisms of VOCs from cellular immunity are not well elaborated. In this review, we discussed how SARS-CoV-2 VOCs adapt inside host cells and escape anti-COVID-19 cellular immunity, focusing on the effect of specific SARS-CoV-2 mutations in hampering the activation of CD8+ T-cell immunity.
The current pandemic of coronavirus disease 2019 (COVID-19) drives the global population in a deep phobia, as the COVID-19–associated burden is critical, resulting in thousands of deaths each day. As of 09 August 2022, there have been 590,443,154 people infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus responsible for COVID-19, with 6,439,059 deaths worldwide. Around 21,787,511 people are still actively infected, and approximately 0.5 million new cases and 658 new deaths are reported daily, with a re-increasing trend of new infections observed from the beginning of the 2021 winter (https://covid19.who.int/; https://www.worldometers.info/coronavirus/). Several studies have reported that these new COVID-19 cases (or waves) are more likely to be caused by infections with emerging SARS-CoV-2 variants of concern (VOCs) (in 98% of cases) than infections with the wild-type (WT) SARS-CoV-2 strain (1–3) initially isolated in Wuhan, China in December 2019 (4, 5). Based on the Pango nomenclature system (6–9), the WHO and the CDC defined VOCs as “variants associated with a high degree of transmissibility, disease severity, neutralizing antibody and vaccine resistance, reduced treatment effectiveness, or diagnostic detection failure” (https://www.cdc.gov/coronavirus/2019-ncov/variants/variant-classifications.html#anchor_1632154493691). Indeed, people infected with emerging SARS-CoV-2 VOCs are more infectious than those infected with WT SARS-CoV-2 and other variants, suggesting that the VOCs have a higher ability to spread than the original SARS-CoV-2 strain (1, 3, 10). For instance, Daniloski et al. demonstrated that SARS-CoV-2 mutants bearing only the D614G mutation confer an increased ability to spread more quickly than the WT SARS-CoV-2 (11). Moreover, Chen et al. showed that the Omicron variant might be 10 times more infectious than the WT virus and almost three times as infectious as the Delta variant (12). Moreover, numerous reports support the fact that emerging VOCs are more severe, with a higher mortality risk than WT SARS-CoV-2, can resist prevention and treatment strategies used so far against WT SARS-CoV-2, and can escape preexisting WT SARS-CoV-2 immunity (3, 13–15). Beta, Gamma, and Omicron variants, for instance, have been shown to have reduced neutralization by monoclonal antibody therapy (including bamlanivimab and the Rockefeller University antibody C144 for Omicron), convalescent plasma (CP), and postvaccination sera (3, 16–18). Immune escape by emerging SARS-CoV-2 VOCs is, therefore, the main concern in the COVID-19 pandemic management (19). Mutations occurring in SARS-CoV-2 spike protein may confer on VOCs the ability to adapt to and escape from natural and vaccine-induced immunity and fast spread, resulting in a detrimental effect on public health. Liu et al. (14) reported that mutations in the receptor binding domain (RBD) and N-terminal domain (NTD) play a crucial role in variant resistance to humoral immunity. In particular, mutation at residue S477 found in the Omicron variant confers resistance to CPs, while mutation at residue E484 found in Beta, Gamma, and Omicron variants confers resistance to neutralizing monoclonal antibodies (NmAbs), vaccines, and postvaccine sera (12, 14). The spike mutation at residue K417 in almost all VOCs, but not the Alpha variant, has been predicted to cause an overwhelmingly disruptive effect, which may make these variants resistant to vaccine-induced humoral immunity (3, 12). Overall, at the molecular level, these spike mutations induce molecular tridimensional changes at the antibody binding sites, which become inaccessible for antibodies and therefore impede antibody binding (20, 21). Also, the residues changed are such that mutations induce an increased binding affinity (or interaction force) of RBD to angiotensin-converting enzyme 2 (ACE2), like in the cases of mutations V367F, L452Q, N501Y, and D614G, which is associated with increased transmissibility (1, 10, 11, 21–23). Similarly, antiviral T-cell immunity evasion by VOCs has also been associated with mutations occurring in the SARS-CoV-2 spike protein. More specifically, mutations in several HLA-I-restricted SARS-CoV-2 epitopes were found to effectively allow VOCs, including Alpha, Beta, and Delta, to escape from viral clearance by CD8+ cytotoxic T lymphocytes (CD8+ CTLs) (24–27). For instance, mutations L452R and Y453F found in B.1.427/429 (also known as CAL.20C) and B1.1.298 variants are associated with resistance to cellular immunity (24). Moreover, infection with SARS-CoV-2 VOCs is followed by a decreased production of IFN-γ and CD8+ T-cells and, more interestingly, an almost zero cytotoxic activity of the low titer of CD8+ T-cells produced (25, 28). Also, Le Bert et al. (29) found that in SARS-CoV-2 VOC infections, the cytotoxic activity of CD8+ T-cells inversely correlates with COVID-19 severity, suggesting that mutations in SARS-CoV-2 S protein may affect the functionality of CD8+ T-cell immune response. More specifically, they may probably induce mechanisms inhibiting the cytotoxic activity of CD8+ T-cells (25, 26), which therefore allow their over-replication and spread. Unfortunately, unlike the well-documented detrimental effect of mutations on humoral immunity, how the mutations in SARS-CoV-2 VOCs induce T-cell immunity evasion at the molecular level is not well documented. In this review, we discussed how SARS-CoV-2 VOCs adapt to and escape from anti-COVID-19 cellular immunity by focusing on the effects of specific SARS-CoV-2 mutations on cytotoxic CD8+ T-cell immunity activation.
Most acute respiratory viral infections trigger activation and proliferation of both naïve CD4+ and CD8+ T-cells, as they play central roles in viral clearance. For instance, mature effector CD8+ CTLs are known to block virus multiplication by killing infected cells and secrete antiviral cytokines, including IFN-γ, TNF-α, and infected-cell killer molecules [Fas-L, perforin, and granzyme B (GrB)] (30–32). The molecular mechanism for activating naïve CD8+ T-cells consists of two main pathways, namely thymus-independent (33) and thymus-dependent activation pathways (34, 35). In the thymus-independent activation pathway, CD8+ T-cell activation requires virus-infected antigen-presenting cells (APCs), which present a cognate viral peptide to naïve CD8+ T-cells. Specifically, following viral entry, the proteasome and other peptidases in the cytosol progressively degrade viral proteins to small specific peptides. The generated peptides are transported into the endoplasmic reticulum (ER) and trimmed by ER aminopeptidase 1, and those with the appropriate/specific motif are loaded onto MHC I molecules. Through the Golgi, peptide-MHC-I molecule complexes transit the plasma membrane and display the loaded viral antigen at the APC surface. Thus, CD8+ T-cell activation occurs when the T-cell receptors (TCRs) of CD8+ T lymphocytes recognize viral peptides loaded onto MHC I molecules [reviewed in (31)]. Moreover, in the absence of virus-infected APCs displaying their cognate peptide through MHC I molecule binding to naïve TCR CD8+ T-cells in secondary lymphoid organs (lymph nodes and spleen), induction of CD8+ cytotoxic T lymphocytes may require help from active CD4+ T helper cells (31, 34, 35). In this activation pathway, two models have been described: the two- and three-cell models. In the former model, CD4+ T-cells first pre-activate APCs such as dendritic cells (DCs) by co-stimulation, which subsequently activate naïve CD8+ T-cells. In the later model, both active CD4+ Th and naïve CD8+ T-cells interact simultaneously with the same APC, and naïve CD8+ T-cell activation occurs through interleukin-2 (IL-2) production by CD4+ Th cells (31, 34–37). After activation, specific mechanisms regulating differentiation and determining the fate of effector CD8+ T-cells occur [reviewed in (38)]. Overall, most (but not all) effector CD8+ T-cells expand and differentiate into mature effector CTLs to clear viral infections. After viral clearance, the mature effectors that have a shortened lifespan die, while the small remaining set of activated CD8+ T-cells differentiates into memory CD8+ T-cells, which will help to control secondary infections more efficiently and rapidly (38).
Studies on T-cell immune responses to SARS-CoV-2 infection are scarce. The substantial role of cellular immunity in SARS-CoV-2 infection has been demonstrated in the few available agammaglobulinemia-related studies, where a standalone T-cell response could complete the viral clearance and assure full recovery in humoral immunodeficiency patients (39–42). Therefore, in COVID-19, like in other respiratory diseases, SARS-CoV-2 infection is followed by a huge and robust immune response mediated by a variety of T-cells, phenotypically and functionally diverse, protecting from severe complications, leading to a quick recovery and conferring long-lasting (memory) immunity. More specifically, in symptomatic and acute COVID-19 patients, clinical reports have shown a state of characterized lymphopenia, especially in moderate-to-critically ill COVID-19 patients (43–47), in which the T-cell count was lower than that in mild COVID-19 patients and healthy people (normal range 955–2,860 T-cells/µl) (25, 28, 48, 49). Moreover, given that the elderly infected with SARS-CoV-2 have the worst disease outcomes (50), leading to death (51), aged-based studies showed that cellular immune response is reduced, and the T-cell count is far lower in the elderly than that in healthy donors and mild and recovered patients (52). In contrast, in mild COVID-19 patients, a higher T-cell response was observed and characterized in almost all patients (detection of CD4+ and CD8+ in 80–100% and 70%–80% of COVID-19 patients, respectively [reviewed in (53)], with a higher CD8+/CD4+ T-cell ratio, along with a higher T-cell count than neutrophils (54, 55). Also, in convalescent antibody-positive and -negative COVID-19 patients, a robust T-cell response was characterized by the presence of reactive CD4+CD154+CD137+ and CD154+CD137+ T-cells (41). Moreover, other T-cells with activated phenotypes, including CD38+, CD39+, HLA-DR+, Ki-67+, and CD69+ T-cells, were detected mostly in mild and convalescent COVID-19 patients [reviewed in (53)]. These observations, which positively correlated with the clearance of COVID-19 symptoms and recovery of almost all patients without artificial respiratory assistance, were significantly opposite to those observed in moderate and severe COVID-19 patients (53–55). Thus, it is worthy to conclude that the lymphopenia state positively correlates with COVID-19–associated death (i.e., lymphopenia is a death-determining factor) because people who succumbed to COVID-19 had a significantly lower absolute number of lymphocytes (specifically CD4+ and CD8+ T-cells) than convalescent patients (56–58). This indicates that COVID-19 patients with a decreased T-cell response, including CD4+ and CD8+ T-cells, are likely to be more vulnerable to disease severity and fatality, highlighting the central role of CD4+ and CD8+ T-cells in SARS-CoV-2 clearance. Nevertheless, T-cell exhaustion and dysregulation have been described in COVID-19 [reviewed in (53, 56)], mainly at higher viral loads. However, in immunocompetent patients, this condition may be transient, with the return of CD8+ T-cells boosted by effector CD4+ and memory CD4+ T-cells within 2 to 3 months, as observed in SARS-CoV infections (59). Furthermore, the diversity of T-cell response has been associated with the production of abundant protective CTL- and Th1-response–inducing cytokines (60). In convalescent mild and severe COVID-19 patients, a high production frequency of double- and triple-positive IFN-γ–, TNF-α–, and IL-2–producing CD4+ T-cells has been detected. Also, a similar expression of IFN-γ, TNF-α, GrB, and/or the CD107a marker of degranulation producing CD8+ T-cells has been reported (41, 49, 54, 56, 61, 62). In that view, Jordan et al. (63) specified that IL-2 and TNF-α are markers for activated CD4+ T-cells and TNF-α and IFN-γ for activated CD8+ T-cells. In more severe COVID-19 cases, however, elevated and steady exhaustion levels and reduced functional diversity of T-cells in peripheral blood together with higher production levels of type 2 (IL-5, IL-9, IL-10, and IL-13) and type 3 (IL-17A/F and IL-22) responses have been found, suggesting that this later promotes the activation of the production of proinflammatory cytokines, including IL-1β, IL-6, CXCL8/IL-8, TNF, and CXCL10/IP-10, also associated with neutrophils and lymphoid organ damage (blocking T-cell response) (61, 64, 65), in severe COVID-19 patients. The detectable reactive T-cell response in COVID-19 patients responsible for the viral clearance has a broad variable specificity to different SARS-CoV-2 proteins. The most dominant reactive T-cells, including CD4+, CD8+, CD4+CD154+CD137+, and CD154+CD137+, detected in mild and recovered COVID-19 patients were specific to SARS-CoV-2 structural proteins (SPs), including ORF3a, spike (S), membrane (M), and nucleocapsid (N) (41, 53, 66). Non-structural protein (NSP)-specific T-cells, including SARS-CoV-2 NSP13 of ORF-1, NSP7, and ORF7/8, have also been identified (53, 66, 67). Moreover, it is important to mention the existence of cross-reactive cellular immunity. Indeed, several reports demonstrated a preexisting protective T-cell immunity against COVID-19, specific to SP and NSP from human coronaviruses (hCoVs) other than SARS-CoV-2, in healthy and SARS-CoV-2 non-exposed adults and in blood samples obtained before the COVID-19 outbreak. Similarly, SARS-CoV-2–specific T-cell response is found to cross-react with other HCoV proteins (41, 66). This suggests that, similar to SARS-CoV–specific T-cell response, which displays a robust cross-reactivity to SARS-CoV-2 proteins after 17 years post-infection, SARS-CoV-2–specific T-cell immunity may persist in recovered COVID-19 patients, allowing for rapid clearance of the infection in the case of secondary infection with SARS-CoV-2 (66) and—probably—SARS-CoV-2 variants, but not all (68). For instance, some studies reported that the reinfection rate by WT SARS-CoV-2 was very low (absolute rate of 0%–1.1%) in individuals who recovered from WT SARS-CoV-2 infection, and their immune responses were elevated and steady for at least 10 months (68, 69). Note that this estimated law reinfection rate was related to reinfection by the same WT SARS-CoV-2. The low reinfection rate by VOCs due to cross-protection by SARS-CoV-2 T-cell immunity remains speculative and confirmed, even though the preexisting WT SARS-CoV-2 cellular immunity may contribute to the attenuation of VOC-associated clinical severity (68). In contrast, current newborns and children are unlikely to have preexisting cross-reactive T-cell immunity against SARS-CoV-2, as they have not been exposed to SARS, MERS, and/or other circulating HCoVs. This is supported by Cohen et al. (70), who demonstrated that memory CD4+ T-cell response increases with age, and CD8+ T-cell response increases with time post-infection, explaining the significantly lower SARS-CoV-2 T-cell response and preexisting cross-reactive CD4+ and specifically CD8+ T-cell immunity against SARS-CoV-2 in children and newborns than in adults (70). This suggests that CD8+ T-cell immunity will take longer to maturate and clear SARS-CoV-2 infection in infants than in adults.
In a recent study, Alison et al. (71) demonstrated that WT SARS-CoV-2–specific T-cell natural and vaccine-induced immunity is not negatively or is lightly affected by but could still recognize VOCs, including Alpha (B.1.1.7), Beta (B.1.351), Gamma (P.1), and CAL.20C variants, and that only 7% and 3% of CD4+ and CD8+ T-cell epitopes are mutated, respectively. Mazzoni et al. (72) also supported and specified in their study that the WT SARS-CoV-2–specific CD4+ T-cell response is more conserved against VOCs because mutations mainly occur within non-CD4+ T-cell epitopes, which might suggest that clearance of VOC infection could be mediated mainly by preexisting SARS-CoV-2–specific CD4+ T-cells. This allows them and some other scientists (73) to hypothesize that despite mutations in T-cell epitopes and because of the broad conserved T-cell epitope coverage, WT SARS-CoV-2–specific T-cell immune response (regardless of the immunity-mediating T-cell types) may still contribute to reducing SARS-CoV-2 (including WT and VOCs) infection severity. However, mutations in 3% of CD8+ T-cell epitopes make a huge difference. They may lead to indescribable fatalities due to more virulent mutants, as reported by Elisa Guo and Hailong Guo (74). They found that “CD8+ T-cell epitope mutants of SARS-CoV-2 proteins lead to persistently variable SARS-CoV-2 infections with different susceptibility and severity” (74, 75). Indeed, several other studies demonstrated with solid evidence that, despite the preexisting SARS-CoV-2–specific cellular immunity in COVID-19 recovered patients, the viral replication rate after reinfection with SARS-CoV-2, but specifically SARS-CoV-2 VOCs, is increased in these patients (76). More importantly, although the presence of CD8+ T-cell immune response against VOCs in WT COVID-19 convalescent or recovered patients was reported, as claimed previously (71, 72), these CD8+ CTLs were non-functional or ineffective against VOCs (76–78). Gallagher et al. also demonstrated that VOCs escape from vaccine CD8+ T-cell immune response as they found a decreased T-cell immunity against VOCs (Alpha (B.1.1.7), Beta (B.1.351), and B.1.1.248 variants) in patients vaccinated with specific SARS-CoV-2 mRNA vaccines from Moderna and Pfizer compared with T-cell responses to WT SARS-CoV-2 infection (79). These clinical features in COVID-19 suggest that preexisting SARS-CoV-2–specific T-cell responses might be ineffective against infection with VOCs and imply that SARS-CoV-2, but more probably VOCs, can still escape from CD8+ T-cell immunity and lead to inactivation of T-cell immunity while maintaining active viral replication (80–83). This is the main clinical characteristic of the Omicron variant, mainly described as mild symptomatic infection, with an increased infection rate in SARS-CoV-2 recovered patients (83, 84). Furthermore, compared with CD4+ T-cell epitopes, CD8+ T-cell epitopes are more vulnerable. Indeed, CD8+ T-cell HLA-I epitopes are shorter (8 to 10 residues) than CD4+ T-cell HLA-II epitopes (12 to 16 residues). A single mutation in one of the CD8+ T-cell HLA epitopes is enough and sufficient to impair and compromise recognition of epitopes by HLA, thus inhibiting activation, functionality, and cytotoxic activity of CD8+ T-cells, which considerably and specifically inhibits the destruction of infected host cells (62, 75) and generally affects the overall T-cell response efficacy. Understandably, subversion of CD8+ T-cell response affects the potency of the whole T-cell response because, in the context of the SARS-CoV-2 threat, the viral replication mechanism is exclusively intracellular, and the main involved T-cell response is led by CD8+ CTLs, due to efficient presentation of endogenously produced antigens on MHC-I molecules. Pretti et al. (85) demonstrated in an in silico analysis of VOCs’ epitopes of CD8+ T-cells that a single mutation including E484K in spike protein may induce T-cell evasion as it alters the binding of the peptide onto its corresponding HLA of MHC-I ( Table 1 ). More interestingly, it has been shown that non-functional and/or exhaustion of CD8+ T-cells in convalescent non-human primates significantly decreases the protective efficacy of natural immunity against SARS-CoV-2 and promotes infectivity and severity of SARS-CoV-2 VOCs. Also, in critically ill COVID-19 patients, a lower CD8+/CD4+ T-cell ratio was discovered (i.e., a low titer of CD8+ T-cells), suggesting that functional CD8+ T-cells, but better associated with CD4+ T-cells in SARS-CoV-2 infection, are therefore required for preventing infection severity associated with a better viral clearance (24, 25, 28, 29, 53, 88). Prior-to-SARS-CoV-2 outbreak studies demonstrated that antiviral cellular immunity evasion by variants is associated with mutations occurring in CTL epitopes (involved in T-cell activation), which results in enhanced infection severity (89, 90) ( Tables 1 , 2 ). Similarly, recent studies corroborate these previous findings, demonstrating that in infections with emerging SARS-CoV-2 VOCs, there is low production of IFN-γ and CD8+ T-cells and an almost zero cytotoxic activity of the latter (25). Specifically, they demonstrated that non-synonymous single mutations of CD8+ T-cell epitopes found in most VOCs induce inhibition of MHC-I binding in a cell-free in vitro assay, resulting in reduced and non-functional CD8+ T-cell production (25, 26), which demonstrated that mutations in VOCs evade CD8+ T-cell immunity and adapt into host cells ( Table 1 ). The same results were found by Motozono et al. (24), describing a reduced potency of CTL, followed by increased COVID-19 infectivity and severity, in SARS-CoV-2 VOC-infected people. Given the demonstrated negative effect of SARS-CoV-2 mutants on the functionality of CD8+ T-cell immune responses, potential mechanisms underlying these effects must be documented.
In general, viral replication is a natural survival process that viruses go through and which unfortunately causes damage to their hosts, which, in turn, counterattacks to eliminate the viral infection via a protective immune response. To escape the host immunity, especially the cellular but CD8+ T-cell immune response, in COVID-19, SARS-CoV-2 uses certain evasion mechanisms, including genomic changes, under the host immune pressure, which yield variants with selective and survival advantages and enhanced viral fitness. These are literally followed by increased infectivity and severity. These modifications include up- or downregulation of certain viral gene expression mechanisms or non-synonymous mutations in gene sequences involved in immune response activation.
The SARS-CoV-2 ORF8 protein is 121 amino acids long and consists of a covalent disulfide-linked dimer formed through the N-terminal sequence and a separate non-covalent interface formed by 73YIDI76, another SARS-CoV-2–specific sequence. Moreover, the ORF8 protein N-terminal sequence is followed by an Ig-like fold and a signal peptide for endoplasmic reticulum (ER) entry, where ORF8 protein interacts with host proteins, including factors involved in ER-associated degradation (93, 94). It has been found that SARS-CoV-2 uses the product of its ORF8 gene to escape CD8+ T-cell immunity through disruption or a downregulation of the mechanism of antigen presentation to CD8+ T-cells by the MHC-I (82). Specifically, the ORF8 protein of SARS-CoV-2 directly interacts with the MHC-I molecules and strictly induces their downregulation. The direct interaction occurs in the ER, and once the complex ORF8-MHC-I molecule is formed, the ORF8 product induces MHC-I trafficking from the ER to lysosomes mediated by ER-phagy for lysosomal vesicle degradation by autophagy. It is, in fact, the subsequent interaction of ORF8 protein with Beclin 1 [a key molecule in autophagy initiation (95)] that induces activation of the autophagy pathway and the further degradation of MHC-I, which is responsible for the lower sensitivity of SARS-CoV-2–infected cells to lysis by CTLs (82) ( Figure 1 ). This evasion mechanism is enhanced in infections by VOCs (82, 96, 97). Indeed, mutations in the ORF8 gene have been associated with increased severity, transmissibility, and especially immune evasion (86, 94, 96, 98). Specifically, many reports have identified non-synonymous mutations or truncations in the ORF8 gene of VOCs (86, 96), explaining in part the enhanced immune escape by these VOCs, including the variant Alpha (202012/01 or B.1.1.7), which has a mutation (Q27 stop codon) that truncates ORF8 (86). Therefore, these SARS-CoV-2 VOCs use their selective ORF8 mutant proteins to enhance the above-described mechanism of activation of the autophagy pathway and the lysosomal degradation of MHC-I, which yields an increased inactivation of the CTL response ( Figure 1 ). Fortunately, experiments have demonstrated that a knockdown or a complete deletion of ORF8 activates surface MHC-I proper expression and significantly reduces immune escape (82, 96), suggesting that inhibiting ORF8 of SARS-CoV-2 by some specific body-harmless nanoparticles or nanobodies (82) constitutes a way to alleviate immune escape by VOCs and enhance CD8+ T-cell efficacy.
Numerous reports demonstrate that SARS-CoV-2 uses mutation-based strategies to downregulate activation pathways of CD8+ T response and evade viral clearance. Thus, despite the high rate of conserved T-cell epitopes in SARS-CoV-2 mutants (71, 72), any changes occurring in dominant CD8+ CTL epitopes involved in the activation of the T-cell immune response have a negative effect on CD8+ T-cell activation, specifically causing deficiency of antigen HLA-A binding and CD8+ CTL activation (75, 89, 90) ( Figures 2 , 3 ; Table 2 ). Pretti et al. (85) demonstrated that in an in silico analysis of VOCs’ epitopes of CD8+ T-cells, a single mutation including E484K in spike protein induces T-cell evasion as it alters the binding of the peptide onto its corresponding HLA molecules of MHC-I. Qiu et al. (75) also demonstrated that, while dominant CD8+ T-cell epitopes including n-Sp1 of SARS-CoV-2 induce epitope-specific T-cell responses with cytolytic activity toward target cells through HLA-A*02:01 binding, mutations in these epitopes cause potential peptide–HLA-A2 binding deficiency and a decreased CTL activation ( Figures 2 – 4 ). Specifically, of the 15 predicted HLA-A*02:01-restricted peptides of S protein, 13 peptides could bind to HLA-A*02:01, while tetramers from seven peptides (n-Sp1, n-Sp2, n-Sp6, n-Sp7, n-Sp11, n-Sp13, and n-Sp14) could detect antigen-specific CD8+ T-cells in COVID-19 convalescent patients and activate CD8+ T-cell immunity. Subsequent analyses demonstrated that these seven antigen peptides are the least conserved in SARS-CoV-2 variants, bearing 19, 9, 13, 10, 12, 10, and 9 types of variations, respectively, and that these variant peptides hamper the HLA molecule binding and significantly reduce MHC-I antigen presentation and thus CD8+ T-cell activation. This suggests that mutations occur in high frequency in around 50% of CD8+ T-cell epitopes (7/14), reducing CD8+ T-cell activation by half. From a molecular point of view, Zhang et al. (87) recently solved crystal structures of two novel crucial CD8+ T-cell epitopes of SARS-CoV-2 (KIA_S and NYN_S) involved in cellular immunity activation in complex with their HLA molecule receptors (HLA-A*02:01 and HLA-A*24:02, respectively). They showed that KIA_S and NYN_S peptides specifically form strong and stable complexes with HLA-A*02:01 and HLA-A*24:02, respectively ( Figures 2A, B ), which aligns with their respective ability to activate CD8+ T-cell immunity. However, non-synonymous substitutions of residues K417 (from KIA_S) ( Figure 2A ) and L452 (from NYN_S) ( Figure 2B ), which are not conserved in either of the three VOCs (B.1.1.7, B.1.351, or P.1), lead to the loss of affinity of the two mutant peptides to their specific relevant HLA and significantly induce relative VOCs to prevent the activation of and escape from CD8+ CTL responses (87). More specifically, in the KIA_S/HLA-A*02:01 complex, the cation–pi interaction (K417–W167 bound) is the main bond that stabilizes the complex ( Figure 2A ) (87) over others (salt bridge interactions), which are weakened due to the acidic environment in the Golgi (99). In VOCs, including B.1.1.7, B.1.351, and P.1 lineages, this highly positively charged residue (K417) is changed by chargeless residues (Asp or Thr) ( Table 2 ), which abolish the cation–pi interaction, yielding low HLA-binding affinity. Similarly, in the NYN_S/HLA-A*24:02 complex, L452 mediating hydrophobic interactions is primarily responsible for the high-affinity binding and stabilization of this complex ( Figure 2B ), despite the presence of salt bridge interactions (99–101). The non-silence mutation of leucine to arginine in VOCs abolishes the hydrophobic interactions, resulting in a loss of affinity for HLA. Overall, mutated peptides cannot be loaded onto their respective HLA molecules and presented by MHC-I to CD8+ T-cells, resulting in the inactivation of cytotoxic responses (CD8+ CTLs).
Wu et al. (27) solved two CD8+ T-cell epitope structures in complex with HLA-A2 (RLQ–HLA-A*02:01 and YLQ–HLA-A*02:01) and with their respective TCRs (RLQ3–RLQ–HLA-A*02:01 and YLQ7–YLQ–HLA-A*02:01). As discussed previously, the wild-type RLQ and YLQ peptides form strong and stable complexes, mainly stabilized by Leu-1001 and Thr-274, respectively ( Figures 2C, D ). Similarly, RLQ3 and YLQ7 TCRs form strong and stable complexes with RLQ–HLA-A*02:01 and YLQ–HLA-A*02:01, respectively, featured by Arg-1000, Ser-1003, Leu-1004, Gln-1005, Thr-1006, and Tyr-1007 for the RLQ3–RLQ–HLA-A*02:01 complex and Tyr-269, Pro-272, Arg-273, Thr-274, Phe-275, and Leu-277 for the YLQ7–YLQ–HLA-A*02:01 complex ( Figures 2C, D , 3 ), which mediate binding with TCRs. These structural characteristics of the HLA–peptide–TCR complexes perfectly align with the respective ability of TCRs to interact with HLA peptides and activate CD8+ T-cell responses. Interestingly, TCR RLQ3 and YLQ7 could not recognize homologous RLQs and YLQs from other sarbecoviruses, nor could they recognize dominant SARS-CoV-2 RLQ and YLQ peptide variants and induce a CD8+ T-cell response. The most dominant variants in the SARS-CoV-2 VOCs include Q1005H and T1006I for RLQ, and L270F and P272L for YLQ ( Table 2 ). Thus, it was evidenced that mutants T1006I and L270F (25) drastically reduce the binding affinity of RLQ and YLQ to and their loading onto HLA-A2s, as the stabilized interactions mediated by T1006 in RLQ–HLA-A2 and L270 in YLQ–HLA-A2 are abolished ( Figures 2C, D , 3 ). In HLA–peptide–TCR complexes, mutation T1006I impairs HLA-A2–RLQ recognition by TCR RLQ3 because, together with Gln-1005, Thr-1006 are principal stabilizers of RLQ3–RLQ–HLA-A*02:01 complex as they establish the strongest bonds, including hydrogen and van der Waals interactions in the structure (27) ( Figures 3A, B ). Similarly, in the YLQ7–YLQ–HLA-A*02:01 structure, Arg-273 and Thr-274 form the most and strongest contacts (38/62 van der Waals and 14/15 contacts) with YLQ7 ( Figures 3B–E ); thus, mutations in one or both of these residues completely disrupt the recognition of YLQ–HLA-A*02:01 by YLQ7. Taken together, these mutations disrupt not only epitope binding to HLA-A2 but also and especially HLA-A2–epitope binding to TCRs, which corroborates the inability of mutated RLQ and YLQ to activate CD8+ T-cell responses. This phenomenon of selective mutations at specific antigenic sites or at CD8+ T-cell epitopes aiming to reduce affinity to HLA molecules and TCRs, demonstrated for these four amino acids, is commonly used by all VOCs to hamper immune response activation and successfully escape from it ( Figure 4 ; Table 2 ). Table 2 presents the validated CD8+ T-cell epitopes for which mutations induce a reduced T-cell immunity against VOCs and virus immune escape. The direct impact of CD8+ T-cell inactivation through mutated CD8+ T-cell-dominant epitopes is the loss of chemotactic mechanisms, allowing the production and accumulation of proinflammatory cytokines and the recruitment of immune cells involved in eliminating the VOC-infected cells (25).
Another mechanism suggested to be adopted by SARS-CoV-2 VOCs to escape the CD8+ T-cell response includes the direct destruction of the T-cells and/or the damage of the lymphoid organs producing T-cells. In fact, during infection, SARS-CoV-2 targets and infects the lymphocytes, which they kill (102), yielding lymphocyte depletion, known as lymphopenia (or lymphocytopenia), which is a common characteristic of COVID-19 severity (102, 103). More interestingly, lymphopenia is also well explained by the fact that SARS-CoV-2 may trigger the production of proinflammatory cytokines, including IL-1β, IL-8, IL-6, CXCL8/IL-8, TNF, and CXCL10/IP-10 in infected macrophages and dendritic cells, which directly decimate lymphoid organs, including spleen, lymph nodes, bone marrow, and thymus, and therefore blocking T-cell (including CD8+ T-cell) activation (61, 64, 65, 102–104). Specifically, postmortem autopsies from spleens of deceased COVID-19 patients showed that CD8+ T-cells were extremely low in all patients, and inflammatory cytokines (IL-6, IL-8, and IL-10) were increased, along with severe spleen tissue damage. Also, necrosis and lymphocyte apoptosis were detected in most patients, whereas artery thrombosis and spleen damage were observed in all patients (103, 104). This suggests that SARS-CoV-2 infection directly damaged the spleen and atrophied lymphoid follicles, yielding low production of CD8+ T-cells and NK cells. Moreover, a positive link has been established between T-cell death (or exhaustion) and an increased expression of immune checkpoint inhibitor proteins (PD-1/PD-L1) at the CD4+ and CD8+ T-cell surface in severe SARS-CoV-2 patients (105, 106). For instance, in SARS-CoV-2 patients, it was demonstrated that overexpression of PD-1 and PD-L1 induces the activation of the PD-1/PD-L1 signaling pathway, which downregulates the activation of effector T-cell responses through a programmed T-cell death mechanism and predicts COVID-19 severity (52, 106). In the study of Ronchi et al. (106), severe COVID-19 patients and patients who died from COVID-19 had a depleted T-cell response, especially CD8+ T-cells, and a high viral load with a hyperexpression of PD-L1 by pneumocytes. This suggests that SARS-CoV-2 infection induces upregulation of a PD-1/PD-L1 signaling pathway, which is responsible for the T-cell death and CD8+ T-cell immune escape. Consequently, the virus gains the advantage of this state being more threatening. These mechanisms might be enhanced in SARS-CoV-2 VOC infection cases given the successful and noteworthy evasion by VOCs of CD8+ T-cell response. Future studies should address the contribution of SARS-CoV-2 VOCs to programmed lymphocyte death and lymphoid organ damage.
Humoral immunity to SARS-CoV-2 is widely studied as it plays an essential role in virus recognition and neutralization through neutralizing antibodies. However, this role is only limited to extracellular environmental scenarios. Moreover, memory antibodies and B-cells are relatively short-lived, non-persistent over the years, and become undetectable after 4 years post-infection (107, 108), compared with T-cells that can last longer and persist for more than a decade (66). Moreover, occasionally and paradoxically, antibodies can increase virus severity through the antibody-dependent enhancement (ADE) phenomenon. These limits would push scientists to focus on cellular immune response, which plays an important role, too, as it is involved in the destruction and eradication of the infected cells carrying virus particles. Therefore, cellular immunity is as essential as humoral immunity in infection clearance. Despite the efforts put toward the development of strategies to fight against SARS-CoV-2 infections and COVID-19, we still have a long way to go because of the emergence of new SARS-CoV-2 variants, including VOCs, which are more virulent and severe than authentic SARS-CoV-2, and especially resistant to CPs, SARS-CoV-2-specific NmAbs, and the current vaccines (13, 14). Remarkably, since the beginning of the outbreak, the pattern of the COVID-19 pandemic shows surges in new cases and fatalities, followed by declines, and as of now, the world faces a new COVID-19 wave since January 2022 that peaked early in February (https://covid19.who.int/; https://www.worldometers.info/coronavirus/). Interestingly, most of these new cases are caused by the new SARS-CoV-2 strains (1–3) classified by the WHO and CDC as VOCs (https://www.cdc.gov/coronavirus/2019-ncov/variants/variant-classifications.html#anchor_1632154493691). Indeed, it is the non-silent mutations occurring in SARS-CoV-2 that confer to VOCs the ability to escape from innate and adaptive immunity, especially from CD8+ T-cell immunity, and exert their virulence in humans (11, 12, 14) ( Table 1 ). While some studies have demonstrated highly conserved CD8+ and CD4+ T-cell epitopes in VOCs (71, 72), with evidence of T-cell response similarities between WT and mutants (63), it is important to note that the few mutational rates present in structural and non-structural gene products of VOCs, specifically within the CD8+ T-cell epitopes, can exert a cellular immune escape, leading to fatalities. Given the small size of CD8+ T-cell epitopes (8 to 10 amino acids), a single mutation within these epitopes is sufficient to disrupt CTL response (62, 85). This needs to be taken into account given the fact that people who present deficient or non-functional (or non-active) CD8+ T-cells, even with stable CD4+ T-cell response, are vulnerable and susceptible to COVID-19 severity (76–78, 80–83, 109). Two shreds of evidence have been presented here: (i) high CD4+ T-cell titers but low CD8+ T-cell titers were found in critically ill COVID-19 patients infected with VOCs, whereas the opposite was found in mild and recovered COVID-19 patients (53–55); (ii) SARS-CoV-2 recovered patients have genetically conserved T-cell immunity, which can also specifically recognize VOCs; however, these patients can still be reinfected by new SARS-CoV-2 VOCs and, more interestingly, they can experience severe acute respiratory distress syndrome (ARDS) (76, 80–83, 109). These suggest that mutations that occurred in WT SARS-CoV-2 leading to VOCs have negative effects on the production of CD8+ T-cells, and VOCs can still escape from SARS-CoV-2–specific preexisting cell immunity (80, 81), which, even at high titers, may not be as effective as it would be if reinfection occurred with the original WT SARS-CoV-2. Among the evasion pathways that SARS-CoV-2 VOCs may adopt to escape from natural and/or vaccine-induced CD8+ T-cell immune responses specific to WT SARS-CoV-2, we summarize three possible mechanisms: • SARS-CoV-2 VOCs also adopt and enhance the SARS-CoV-2 mechanism of activation of the autophagy pathway and the lysosomal degradation of MHC-I, which highly decreases the activation of the CD8+ CTL response due to mutations of the ORF8 ( Figure 1 ) (86, 96); • Mutations in CD8+ T-cell epitopes specific to SARS-CoV-2 proteins induce a loss of affinity and cannot be loaded onto HLA-A molecules, which results in a lack of TCR recognition and cytotoxicity activation ( Figures 2 – 4 ) (27, 80, 81, 87); • SARS-CoV-2 VOCs may induce enhanced direct destruction of the T-cells and/or damage of the lymphoid organs producing T-cells, specifically CD8+ T-cells, through the hyperactivation of the PD-1/PD-L1 signaling pathway. While mutations in SARS-CoV-2 SP and NSP, specifically in CD8+ T-cell epitopes, have been demonstrated to induce VOC immune escape through the inactivation or downregulation of CTL, future studies should address the contribution of SARS-CoV-2 VOCs, especially mutations in CD8+ T-cell epitopes, on programmed lymphocyte death and lymphoid organ damage, specifically in the overexpression of PD-1 and PD-L1 on CD8+ T-cells. From a reverse point of view, considering studies claiming that reported mutations occurring in CD8+ T epitopes have no effects on WT SARS-CoV-2 T-cell response (71, 72, 80, 81), specifically on CD8+ T-cell activation and are barely preserved within VOCs, this could hypothetically imply that these mutations may create new specific VOC CD8+ T-cell epitopes (25, 74, 75, 85), which might contribute to an effective but delayed clearance of VOCs. For instance, recovered patients from WT SARS-CoV-2 infection or WT SARS-CoV-2 vaccinees or both acquired a protective memory T-cell immunity fully against WT SARS-CoV-2 [with a negligible reinfection absolute rate of 0%–1.1% (68)] but more than 50% reduced against new variants (80, 81, 109). In the case of new infection with VOCs, this less than 50% T-cell immunity, especially CD8+ T-cells (80, 81, 109), may not be strong enough to eliminate the new variants in reasonable kinetics as the variant may also escape from preexisting immunity. Thus, hypothetical new CD8+ T-cell epitopes would be loaded onto corresponding HLA-I molecules and trigger new and specific T-cell activation for a complete—delayed—VOC clearance. Studies by Qiu et al. (75) and Elisa Guo and Hailong Guo (74) demonstrated the possibility of new CD8+ T-cell epitopes from mutated epitopes of SARS-CoV-2, with the ability to increase T-cell activation marker CD69 and CD137 and induce low titer CD8+ T-cell response specific to the mutants, but then, no more specific to the WT SARS-CoV-2. Future studies need to assess the possibility of new epitopes from SARS-CoV-2 VOC infections and their effectiveness in the clearance of SARS-CoV-2 VOCs. These studies would help in developing variant-specific vaccines. Additionally, other studies have raised the conclusion that despite mutations occurring in SARS-CoV-2, which are responsible for SARS-CoV-2 emerging variants (including VOCs), recovered WT SARS-CoV-2 individuals and WT SARS-CoV-2–specific vaccinees retain immunity that cross-reacts with new variants and may clear the VOC infections and prevent them from severe forms of COVID-19 (68, 73). However, this—early—immunity effectiveness might be mainly attributed to memory CD4+ T-cells and, to a lesser extent, to memory B-cells and antibodies, but probably not to memory CD8+ T-cells. This is because, as described in Section 4, mutated epitopes carried by VOCs may no longer be recognized by preexisting CD8+ T-cell immunity, as mutations in SARS-CoV-2 negatively affect mainly CD8+ T-cell epitopes that are more vulnerable (62, 75), but not CD4+ T-cell epitopes for which the same preexisting SARS-CoV-2 immune response still retains efficacy against mutants and may appropriately reduce VOC infection-associated severity (72). Consequently, we suggested that the clearance of VOC infections later on without intensive care admission could be possibly attributed to the development of new CD8+ T-cell epitopes specific to variants, together with the conserved preexisting CD4+ T-cell, which aligns with the global pattern of the COVID-19 pandemic [surges in new cases followed by prevalence declines months later (https://covid19.who.int/)]. In conclusion, to block the CTL-mediated cellular immune escape by SARS-CoV-2 VOCs, studies should focus on the development of new vaccines (such as RNA vaccines, which are known to promote the activation of cellular immunity) and especially on how to boost the CD8+ T-cell response against VOCs (110, 111). Besides RNA vaccines, a better alternative for next-generation vaccines includes epitopes-based vaccines. By focusing on non-structural proteins and spike and nucleocapsid protein domains of SARS-CoV-2 that are relatively less mutated or highly conserved, numerous studies demonstrated dominant CD8+ CTL epitopes specific for HLA-A*24:02 and HLA-A*02:01 genotypes, with a relatively low or zero divergence rate, that can be targeted for developing wild spectrum COVID-19 vaccines, effective against any SARS-CoV-2 variants—and extensively against sarbecoviruses—with the ability to induce neutralizing antibodies and activate specific CD8+ CTLs (27, 92, 112–115). For example, considering five randomly evidenced CD8+ CTL-specific epitopes with low/no mutational rates, such as FLNGSCGSV and VLAWLYAAV (91), PDPSKPSKR, DPSKPSKRS, and QTQTNSPRR (113), new-generation epitope-based vaccines might consist of developing a multivalent-epitope–based cocktail against SARS-CoV-2 from these five epitopes, with peptide carriers and/or intramolecular adjuvants. Besides boosting CD8+ T-cell response, one of the most attractive advantages of such multiple epitope-based vaccines includes the ability to reduce the potential of new SARS-CoV-2 emerging variant development. More interestingly, these epitope-based vaccines have more benefits, including time- and cost-effectiveness, maximal therapeutic efficacy (enhanced antigenicity and immunogenicity), and well-tolerability with minimal adverse effects (113, 115, 116). Also, reports have demonstrated that a knockdown or a complete deletion of ORF8 activates surface MHC-I proper expression and significantly reduces immune escape (82, 96), suggesting that inhibiting ORF8 of SARS-CoV-2 constitutes a way to enhance CD8+ T-cell efficacy against SARS-CoV-2 VOC infections.
AK conceived the presented idea, extracted the data, wrote the original draft, and formatted the manuscript for submission. FB and ZN reviewed and edited the final version for publication. TJ conceptualized the main idea, provided resources and financial assistance during the whole study, and supervised the whole paper. All authors contributed to the article and approved the submitted version.
TJ is supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDB29030104), the National Natural Science Fund (Grant No.: 31870731), the Fundamental Research Funds for the Central Universities, and the 100 Talents Programme of The Chinese Academy of Sciences. AK is supported by the Chinese National Postdoctorate Subvention. ZNN is supported by a Chinese government scholarship.
We apologize in advance to colleagues whose work was overlooked because of length limitations or by our own ignorance.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647063 | Karen Giménez-Orenga,Justine Pierquin,Joanna Brunel,Benjamin Charvet,Eva Martín-Martínez,Hervé Perron,Elisa Oltra | HERV-W ENV antigenemia and correlation of increased anti-SARS-CoV-2 immunoglobulin levels with post-COVID-19 symptoms | 27-10-2022 | HERV-W,long COVID-19,post-COVID,SARS-CoV-2,serology,IgE,chronic fatigue syndrome,CFS | Due to the wide scope and persistence of COVID-19´s pandemic, post-COVID-19 condition represents a post-viral syndrome of unprecedented dimensions. SARS-CoV-2, in line with other infectious agents, has the capacity to activate dormant human endogenous retroviral sequences ancestrally integrated in human genomes (HERVs). This activation was shown to relate to aggravated COVID-19 patient´s symptom severity. Despite our limited understanding of how HERVs are turned off upon infection clearance, or how HERVs mediate long-term effects when their transcription remains aberrantly on, the participation of these elements in neurologic disease, such as multiple sclerosis, is already settling the basis for effective therapeutic solutions. These observations support an urgent need to identify the mechanisms that lead to HERV expression with SARS-CoV-2 infection, on the one hand, and to answer whether persistent HERV expression exists in post-COVID-19 condition, on the other. The present study shows, for the first time, that the HERV-W ENV protein can still be actively expressed long after SARS-CoV-2 infection is resolved in post-COVID-19 condition patients. Moreover, increased anti-SARS-CoV-2 immunoglobulins in post-COVID-19 condition, particularly high anti-SARS-CoV-2 immunoglobulin levels of the E isotype (IgE), seem to strongly correlate with deteriorated patient physical function (r=-0.8057, p<0.01). These results indicate that HERV-W ENV antigenemia and anti-SARS-CoV-2 IgE serology should be further studied to better characterize post-COVID-19 condition pathogenic drivers potentially differing in subsets of patients with various symptoms. They also point out that such biomarkers may serve to design therapeutic options for precision medicine in post-COVID-19 condition. | HERV-W ENV antigenemia and correlation of increased anti-SARS-CoV-2 immunoglobulin levels with post-COVID-19 symptoms
Due to the wide scope and persistence of COVID-19´s pandemic, post-COVID-19 condition represents a post-viral syndrome of unprecedented dimensions. SARS-CoV-2, in line with other infectious agents, has the capacity to activate dormant human endogenous retroviral sequences ancestrally integrated in human genomes (HERVs). This activation was shown to relate to aggravated COVID-19 patient´s symptom severity. Despite our limited understanding of how HERVs are turned off upon infection clearance, or how HERVs mediate long-term effects when their transcription remains aberrantly on, the participation of these elements in neurologic disease, such as multiple sclerosis, is already settling the basis for effective therapeutic solutions. These observations support an urgent need to identify the mechanisms that lead to HERV expression with SARS-CoV-2 infection, on the one hand, and to answer whether persistent HERV expression exists in post-COVID-19 condition, on the other. The present study shows, for the first time, that the HERV-W ENV protein can still be actively expressed long after SARS-CoV-2 infection is resolved in post-COVID-19 condition patients. Moreover, increased anti-SARS-CoV-2 immunoglobulins in post-COVID-19 condition, particularly high anti-SARS-CoV-2 immunoglobulin levels of the E isotype (IgE), seem to strongly correlate with deteriorated patient physical function (r=-0.8057, p<0.01). These results indicate that HERV-W ENV antigenemia and anti-SARS-CoV-2 IgE serology should be further studied to better characterize post-COVID-19 condition pathogenic drivers potentially differing in subsets of patients with various symptoms. They also point out that such biomarkers may serve to design therapeutic options for precision medicine in post-COVID-19 condition.
The sanitary and economic burden associating with patient post-viral COVID-19 sequalae, months or even years after virus clearance constitutes a prominent pandemic aftermath challenge, particularly when considering the large number of affected individuals worldwide (1). Although this medical condition was listed in the ICD-10 as “post-COVID-19 condition” since September 2020 (2), the term popularized for SARS-CoV-2 post-viral syndrome whose main persistent symptoms include sore throat, dyspnea, chronic fatigue, pain, intestinal and sleeping disturbances, cognitive deficits, anxiety and depression (3) seems to be “long COVID-19”. Current clinical treatments for post-COVID-19 condition are merely symptom palliative since the potential mechanisms triggering or underlying the syndrome remain undefined and no specific clinical biomarkers have yet been identified (3, 4). A situation shared by other diseases of unknown etiology presenting overlapping symptomatology with post-COVID-19 (5–7). The intense research in this field is, however, starting to yield valuable hints to improve our understanding of post-COVID-19 condition physiologic derangements, as for example the remodeling of T cell dynamics upon SARS-CoV-2 infection (8), the viral-induced autoimmunity (9, 10), or the harm derived from SARS-CoV-2 neuro-invasive capacity (11, 12). Viral infections may trigger epigenetic changes leading to human endogenous retrovirus (HERVs) aberrant expression and chronic innate immune activation which translates into patient post-viral symptoms (13). HERVs are ancient retroviral DNA sequences integrated into the human genome during evolution by reaching the germline, which constitute 8% of its size (14). Despite the benefits provided by their regulatory and coding capacities (14–16), it is widely documented that aberrant HERV expression correlates with neurological disease (14, 17–19). The recent detection of the multiple sclerosis antigen HERV-W ENV in the blood of acute COVID-19 patients, and its demonstrated relationship with disease severity and inflammation (20, 21), raised the question on whether, similarly to Epstein-Barr Virus (EBV) infection and multiple sclerosis (22), SARS-CoV-2 infection may trigger long-term neurologic disease through HERV activation. On another end, the study of antibody response to SARS-CoV-2 by screening virus antigen microarrays revealed that early IgA and IgG responses can constitute markers of acute disease severity (23). In addition, it was found that SARS-CoV-2 elicits IgE responses with levels positively correlating with severity thus suggesting a link between SARS-CoV-2 infection, degree of severity with mast cell activation (23, 24). This study aimed at determining if in fact HERV-W ENV expression remains active in post-COVID-19 patients and, if so, what is the relationship between HERV-W ENV expression and anti-SARS-CoV-2 immunoglobulin levels, both being increased in acute SARS-CoV-2 infections with severity, in reference to patient symptoms. Importantly, the outcome could serve to unravel post-COVID-19 subjacent mechanisms relating to acute SARS-CoV-2 infections. Information holding diagnostic and therapeutic implications.
This cross-sectional study was approved by the Public Health Research Ethics Committee DGSP-CSISP of Valencia, Valencia, Spain, núm. 20210604/04/01. Samples and data from patients included in this study were provided by the IBSP-CV Biobank (PT17/0015/0017), integrated in the Spanish National Biobanks Network and in the Valencian Biobanking Network and they were processed following standard operating procedures with the appropriate approval of the Ethics and Scientific Committees (num. 20210604/04/02). Provision of samples of acute COVID-19 cases from the biobank of “Hospices Civils de Lyon” was approved by the ethical committee (Centre de Resource Biologique de Hospices Civiles de Lyon, Hôpital de la Croix-Rousse, Lyon France) and the French Ministry of Research for the constitution of a collection of COVID-19 biological samples and their session for the purpose of research (Autorisation N°: DC-2020-3919 and AC-2020-3918). Peripheral blood mononuclear cells (PBMC) from healthy donors (HBD) were obtained under established legal activity for the provision of blood-derived samples from the “Etablissement Français du Sang” (EFS) of Lyon (France). The study included a total of 66 patients diagnosed with acute COVID-19 (n=22), post-COVID-19 condition (n=12), chronic fatigue syndrome (CFS) (n=17) -a disease presenting overlapping symptoms with post-COVID-19 condition- (pre-pandemic cases/samples) (7), pre-pandemic healthy blood donors (ppHBD) (n=4) and similar healthy donors with samples obtained during the pandemic (pHBD) (n=11) ( Supplementary Table 1 ). CFS patients were diagnosed using the Canadian (25) and/or International Consensus (26) criteria. Post-COVID-19 condition patients suffered the disease for at least six months. Acute COVID-19 patients were hospitalized with moderate to severe clinical status based on the clinical scale for COVID-19, recommended by the National Institute of Health of the USA guidelines (https://www.covid19treatmentguidelines.nih.gov/overview/clinical-spectrum/). Patients with health problems other than CFS or COVID-19 were excluded from the study and individuals with any similar or related pathology were excluded from HBD groups. Written informed consent was obtained from all study participants, and CFS and post-COVID-19 patient health status was evaluated with the use of standardized questionnaires, including the FIQ case report form (27, 28), the MFI questionnaire (29), and the quality-of-life SF-36 instrument (30).
Each participant donated 10 mls of blood obtained by standard phlebotomy in K2EDTA (BD cat. 366643 or Vacuette cat. 455045) tubes. All blood samples were kept at RT and processed within 4 h. Tubes were centrifuged at 500 xg for 15 min to separate plasma from the remaining blood components, the upper phase was transferred to a clean tube and further centrifuged at 8000 xg for 10 min to remove platelets and debris. Poor platelet plasma was aliquoted in cryovials (0.5-1 ml/tube) and kept frozen at -80°C until use, as previously described (31).
For the detection of HERV-W ENV antigen in plasma and the quantification of its circulating soluble form, all analyses were performed according to the conditions provided in the patent published under ref. WO2019201908 (A1) and entitled “Method for the detection of the soluble hydrophilic oligomeric form of HERV-W envelope protein”, as previously described (32). The specific signal was expressed as the signal to noise (S/N) ratio, where the noise represents the mean+2SD of the background signals yielded by a panel of HBD samples. Both HERV-W ENV antigenemia and SARS-CoV-2 serology were performed by Simple Western technology according to immunoglobulin class. Samples of ppHBDs and CFS patients, were stored over a year in freezers at -80°C. Pandemic and acute and post-COVID-19 condition samples were stored in freezers at -80°C for less than a year. All samples were kept frozen at all times after their initial freezing until use (single freeze/thaw cycle before the immunoassay). Detailed description of for SARS-CoV-2 serological analyses with multiple antigens detection is provided as Supplementary Figures 1 , 2 .
All statistical analyses were performed using Prism (version 8.0; GraphPad Software, La Jolla, California). Continuous data are expressed as means ± SD, as indicated. Normal data distribution was assessed using the Shapiro-Wilk test, assuming p-values > 0.05. Statistical differences were determined using t-student or Mann-Whitney tests, depending on whether the data followed or not normal distributions, respectively. To establish group differences, statistical significance was set at p<0.05.
A total of 66 subjects were evaluated in this study corresponding to ppHBDs, n=4; pHBDs, n=11; pre-pandemic CFS patients, n=17; acute COVID-19 patients, n=22; and post-COVID-19 condition patients, n=12. A summary of Total Fibromyalgia Impact Questionnaire (FIQ) (27, 28), Multi Fatigue Inventory (MFI) general fatigue (29) and quality of life Short-Form-36 Health Survey (SF-36) questionnaire (30) scores for CFS and post-COVID-19 condition patients is shown in Table 1 . Differences between CFS and post-COVID-19 condition were only observed with questionnaire scores for total FIQ (p<0.05), MFI general fatigue (p<0.01), SF-36 general health (p<0.05) and mental health (p<0.05), while all other areas among the 17 tested ( Table 1 ) showed no significant difference, thereby suggesting widely overlapping symptomatology across these two groups of patients. However, questionnaire scores differences indicated more intense pain as measured by total FIQ and worse mental health as determined by the SF-36 instrument for the CFS group, while worse fatigue as assessed by the MFI and worse general health (SF-36) were observed for the post-COVID-19 condition cases ( Table 1 ).
Since previous studies had evidenced increased expression of the HERV-W ENV protein in acute COVID-19 patients in association with disease severity (20, 21), this study set to determine whether the presence of this protein remains high in post-COVID-19 patients. To this end, circulating levels of HERV-W ENV protein were measured in post-COVID-19 plasma in comparison to acute COVID-19, ppHBD, and pre-pandemic CFS subjects by Simple Western technology ( Figure 1 ). The pre-pandemic CFS group was included as an additional control group sharing clinical symptoms with post-COVID-19 condition. The analysis interestingly showed that a higher proportion of post-COVID-19 condition subjects, corresponding to 58% (7 out of 12), presented detectable HERV-W ENV protein as compared to only 41% (9 out of 22) of our acute COVID-19 cohort. Moreover, HERV-W ENV protein levels expressed as signal to noise (S/N) ratios were similar in post-COVID-19 condition patients (1.13-to-2.20) when compared to acute COVID-19 (1.02-to-1.8), but significantly different from all controls in both acute and post-COVID-19 condition cohorts (p<0.01). HERV-W ENV expression in long COVID included cases ranging from 6- to 19-month post-infection ( Supplementary Table 1 ). Surprisingly, 2 out of 17 pre-pandemic CFS subjects were strongly positive (1.93 and 2.18) for HERV-W ENV expression, the significance of which is unknown at present.
To explore the relationship of HERV-W ENV protein expression with patient immune response to SARS-CoV2 infection, we measured the levels of circulating anti-SARS-CoV-2 immunoglobulins (IgG, IgM, IgA and IgE) in acute and post-COVID-19 condition cases with respect to pre-pandemic (ppHBD + CFS) and pandemic (pHBD) control groups (also after vaccination campaign started) by Simple Western technology ( Figure 2 ). The recombinant SARS-CoV-2 antigens present in the kit used to detect corresponding specific antibodies in human serum or plasma comprised the virus nucleocapsid and the spike proteins. As expected, pre-pandemic controls were negative for any of the anti-SARS-CoV-2 immunoglobulins while pandemic controls presented reactive IgG and, surprisingly, IgE to a certain extent for the spike antigen in the absence of reactivity to the nucleocapsid antigen, which should be consistent with vaccination ( Figures 2A, D ). Almost all individuals of this group (pHBD) had values over the cut-off for IgG and IgE against SARS-CoV-2 spike, which were statistically different in comparison to pre-pandemic controls (ppHBD + CFS) (p<0.0001 and p<0.001, respectively; right plots in Figures 2A, D ), while very few IgA values and no IgM level surpassed the set cut-off values (right plots in Figures 2B, C ). However, the levels of anti-SARS-CoV-2 spike and nucleocapsid IgG, IgM, IgA and IgE appeared markedly increased in more than a half of acute COVID-19 cases in comparison to control groups ( Figure 2 ). About 59% (13 out of 22) of the acute cases showed a significant increase of total (anti-spike + anti-nucleocapsid) IgG (p<0.0001), IgA (p<0.0001) and IgE (p<0.01) while increased total IgM were detected in even more cases (72%, 16 out of 22) (p<0.001) ( Figure 2 ). This is consistent with a stage of active infection in which large amounts of IgM against the spike and nucleocapsid antigens are still generated. With respect to post-COVID-19 condition cases, a subgroup corresponding to 58% (7 out of 12) showed increased levels of anti-spike IgG compared to pre-pandemic controls (p<0.0001) ( Figure 2A ), and 42% (5 out of 12) showed increased anti-spike IgA levels (p<0.0001) ( Figure 2C ). These results do not seem to be related to recent vaccination, since subjects with equivalent elapsed time from vaccination present very different levels of those immunoglobulins ( Supplementary Figure 3 ). Regarding IgM, just three post-COVID-19 condition cases showed increased anti-nucleocapsid IgM ( Figure 2B ). Strikingly, a large proportion of post-COVID-19 condition cases had detectable IgE levels against SARS-CoV-2 proteins (75%, 9 out of 12), which were significantly increased against the spike protein when compared to acute cases (p<0.05). Both anti-spike and anti-nucleocapsid IgE antibodies were significantly increased when compared to pre-pandemic controls (p<0.01 and p<0.0001, respectively; Figure 2D ). These results highlight that a subgroup of post-COVID-19 condition patients present a sustained immune response against SARS-CoV-2 antigens, with IgG, IgA and IgE levels similar or even increased when compared to those observed during active COVID-19 infections (23, 24). Immunoglobulin levels measured during acute COVID-19 presented moderate to no correlation with each other: IgG levels ( Figure 3A , left and middle plots) showed a moderate correlation with IgA (r=0.6927, p<0.001), but none with IgE (r=0.1896, p>0.05), whereas showing a weak correlation between IgA and IgE levels (r=0.4590, p<0.05; Figure 3A , right plot). Quite differently, in post-COVID-19 condition, three out of the four immunoglobulins studied appeared strongly correlated ( Figure 3B ): positive correlations were found between IgG and IgA or IgE levels, which were close to 1 (r=0.9537, p<0.0001; and r=0.8985, p<0.0001, respectively; Figure 3B , left and middle plots). Likewise, IgA levels correlated with IgE levels to a high extent (r=0.9070, p<0.0001) ( Figure 3B , right plot). This data supports the existence of a sustained immune response against SARS-CoV-2 antigens, consistent between all immunoglobulin isotypes but unexpectedly skewed towards IgE reactivity, months or even years after acute infection.
To explore the potential relationship of SARS-CoV-2 antigenicity with the aberrant expression of HERV-W ENV protein, the levels of IgG, IgM, IgA and IgE against nucleocapsid or spike SARS-CoV-2 proteins were compared across plasma samples showing positive or negative presence of HERV-W ENV in acute and post-COVID-19 condition cases (n=22 and 12 respectively). Interestingly enough, several anti-SARS-CoV-2 immunoglobulins showed significant differences in accordance to HERV-W ENV antigenemia in the acute COVID-19 group ( Figure 4 ). Samples positive for HERV-W ENV (n=9) had increased IgG (p<0.05), IgM (total: p<0.01, nucleocapsid: p<0.05) and IgE (total: p<0.05, spike: p<0.01) levels when compared to negative samples (n=13) ( Figure 4A ). However, no significant difference in SARS-CoV-2 immunoglobulin levels was found in accordance with HERV-W expression in the post-COVID-19 condition group, which, however, may relate to the low number of samples in this group (total n=12, positive n=7). A heatmap illustrating individual differences of the studied cohort is shown on Figure 4B . Though significant differences in specific immunoglobulin levels were observed in accordance with HERV-W ENV positivity in acute COVID-19 cases ( Figure 4A ), no specific correlation between HERV-W ENV and anti-SARS-CoV-2 immunoglobulins levels was found (data not shown).
To further evaluate the potential co-clustering of anti-SARS-CoV-2 immunoglobulin profiles with post-COVID-19 condition symptoms, we measured correlations between anti-SARS-CoV-2 immunoglobulins and patient scores with FIQ, MFI and SF-36 instruments assessing patient health status ( Figure 5 ). Although no significant correlations were found between the majority of the immunoglobulins and the questionnaire scores (data not shown), a moderate to strong correlation was associated with patient´s physical functioning, a variable assessing physical patient performance in the SF-36 questionnaire. As it can be observed in Figure 5 , higher IgG (rs=-0.6502, p<0.05), IgA (rs=-0.5831, p=0.05) but, markedly, IgE (rs=-0.8057, p<0.01) levels associate with lower physical functioning scores relating to increased patient´s disability.
The initial anecdotal cases of COVID-19 long-haulers, term used to designate patients who developed long-lasting COVID symptoms, such as the famous case of a Portland 47-year-old woman who suffered COVID-triggered fever for almost a year (33), turned into a problem of high global proportions (49% affected over 120 days, as determined by Chen et al., recent systematic review and meta-analysis) (34). Clinicians soon realized that no relationship between severity and post-COVID-19 condition forecasting could be established as mild cases could become chronic while some intensive care patients completely returned to normality after a two-month recovery period (33). Among the patients affected on the long-term, two groups have been clearly differentiated based on organ (particularly lungs) damage or on the absence of evidence for such damage. The latter group being identified as potential developers of a syndrome comparable to myalgic encephalomyelitis/chronic fatigue syndrome (CFS) (7, 35), a severely disabling condition characterized by immuno-metabolic deficiencies and cognitive impairment lasting for many years (25, 26, 36). Thus, CFS previously reported to emerge after Lyme’s disease, mononucleosis, influenza, SARS or other infections (37–39), with an estimated worldwide prevalence of 0.89% (with 1.5 to 2-fold of women over men) with data gathered before 2020 (40), could acquire unprecedented values with catastrophic sanitary and economic consequences. This social awareness has motivated great concern and motivation to predict, resolve and prevent post-COVID-19 condition and CFS. A common factor derived from epigenetic changes triggered by infections in neurologic disease is the activation of quiescent retroviral sequences present in the human genomes (HERVs) (13, 14). The finding that HERV-W ENV protein expression is increased in COVID-19 lymphocytes and the fact that HERV-W ENV levels correlate with pneumonia severity (20) motivated our search to determine whether these sequences remain unsilenced in post-COVID-19 condition, perhaps constituting a potential trigger or enhancer for certain symptoms. Our results ( Figure 1 ) evidence for the first time the presence of the HERV-W ENV protein in about one-half of the analyzed plasma samples from post-COVID-19 condition patients. Despite the limited number of long COVID cases analyzed (n=12), the higher proportion and values for HERV-W ENV expression when compared to acute COVID-19 cases (n=22) (58% and 1.13 to 2.20 vs.,41% and 1.02 to 1.8) suggests a potential association of this HERV-encoded protein with the post-COVID-19 condition or, more likely, with a sub-group of patients with related pathogenic consequences. In cases from the present cohort, HERV-W ENV antigenemia is also seen to persist over one-year post-infection ( Supplementary Table 1 ) suggesting that chronic expression is made possible after COVID-19 as seen to be lifelong in multiple sclerosis. Indeed, long-COVID-19, or post-COVID, obviously represents a heterogeneous nosological entity, despite common or overlapping symptomatology between patients (41–43). A mechanism that may explain or favor the activation of HERV-W ENV expression by SARS-CoV-2 could be promoter binding site enrichment for transcription factors involved in immune responses. This would be consistent with the present findings of a correlation between specific immunoglobulin responses and HERV-W ENV expression in acute COVID-19, as well as with the persisting but peculiar humoral immunity in post-COVID-19 condition. While further overexpression of HERV-W ENV may also result from unspecific activation of NF-kB by different viral responses (44), transcription from specific transposable elements (TEs) may be viral agent- and cell phenotype-specific, according to the differential expression obtained from cell cultures infected with Middle East respiratory syndrome coronavirus (MERS-CoV or MERS), influenza A virus (IAV), respiratory syncytial virus (RSV), human parainfluenza virus type 3 (HPIV3), SARS or SARS-CoV-2 (45). Therefore, the study of specific TE profiles altered by SARS-CoV-2 may also reveal relevant features of the deranged cellular pathways in the immune cells of post-COVID-19 condition patients. Interestingly, the highest HERV-W ENV values were obtained from two pre-pandemic CFS cases analyzed here as control samples (not post-COVID-19 condition but presenting overlapping symptoms) ( Figure 1 ). The CFS results support previous findings showing HERV activation in CFS at the transcriptional level (31, 46) and demand further interrogation of HERV element expression at both transcriptional and protein levels. Whether HERV-W ENV aberrant expression is a common aspect to CFS and post-COVID-19 condition, or perhaps other post-viral syndromes, is unknown at present. Of note, it may also define subgroups with related etiopathogenesis in such clinically-defined conditions, as already found in psychiatric disorders with characterized immune inflammation (32). In an effort to further explore the relationship between the activation of the immune system by SARS-CoV-2 and HERV-W ENV expression the exact same samples tested for HERV-W ENV expression were assayed for SARS-CoV-2 multi-isotypes serology (as total anti-SARS-CoV-2 antigens or directed to either SARS-CoV-2 nucleocapsid or spike proteins). Figure 2 shows that SARS-CoV-2 antibody response is increased in acute COVID-19 cases but, rather unusually, for all isotypes (IgG, IgM, IgA and IgE); post-COVID-19 condition cases still had significant levels of anti-SARS-CoV-2 spike IgG, IgA and remarkably above acute cases, IgE isotype, but not of the IgM isotype. The absence of IgM is consistent with a post-acute period, but individual cases still had anti-N SARS-CoV-2 IgM, which should be considered on larger series since gastrointestinal persistence of SARS-CoV-2 infection has now been reported in long COVID cases (47). This also appears consistent with the still elevated anti-spike IgA levels in a significant number of patients with long COVID. Globally, the levels of anti-SARS-CoV-2 spike IgG and IgA observed in the post-COVID-19 condition group are significantly beyond those observed in pre-pandemic healthy controls and correlated to each other as well as to also significantly elevated IgE levels ( Figure 3B ). Two parameters appear to emerge from this pilot study, beyond a mere interest as biomarkers, as potential contributors in the pathogenic pathways involved in long/post-COVID: (i) HERV-W ENV, known to be a potent TLR4 agonist in immunopathogenic as well as in neurotoxic processes (48–50), and (ii) IgE antibodies known to be strong effectors of immunoallergic conditions as well as anti-parasite humoral responses (51, 52). Thus, the present study strongly suggests focussing further studies on their potential role in certain forms and/or symptoms of post-COVID syndrome. Our post-COVID-19 condition patients also presented increased anti-nucleocapsid IgE antibodies. Although the significance of this finding is unknown at present, it seems worth mentioning the correlation found by Matyushkina et al. between anti-SARS-CoV-2 nucleocapsid reactivity and autoimmunity in long COVID patients (10). COVID-19 vaccination intends to boost anti-SARS-CoV-2 spike antibodies (53). In fact, induction of IgG and IgA have been observed with COVID-19 mRNA vaccination with similar induction kinetics, but faster IgA decline after 100 days (54). The absence of IgM anti-SARS-CoV-2 spike protein in control and post-COVID-19 condition participants could be easily explained in the absence of recent infection or vaccination in this group ( Supplementary Table 1 ). To further determine whether SARS-CoV-2 serological profile holds any relationship with HERV expression in acute or post-COVID-19 condition cases, we stratified patient´s according to HERV-W ENV positive or negative expression, and then compared SARS-CoV-2 antibody responses in each subgroup. All anti-SARS-CoV-2 immunoglobulins analyzed showed increased levels with HERV-W ENV expression in acute COVID-19 cases, with the exception of IgA ( Figure 4 ), an observation that may explain the more severe evolution and symptoms already reported in this subgroup of acute COVID-19 patients (20, 21). No differential SARS-CoV-2 IgG, IgM and IgA profile could be attributed to the presence of HERV-W ENV in post-COVID-19 condition but a positive trend was observed for anti-S IgE. Whatsoever, the low number of post-COVID-19 condition cases positive for HERV-W ENV (n=7) in this small pilot cohort does not allow to perform accurate statistical comparisons. Finally, one aim of this analysis of anti-SARS-CoV-2 humoral immunity was to determine whether a sustained immune response or a given isotype profile could characterize post-COVID-19 condition patients. Interestingly, the physical functioning scores, as measured with the standardized, validated, SF-36 instrument (30) appeared significantly correlated with higher Ig levels, particularly IgE levels (rs=-0.8057, p<0.01), meaning that these post-COVID-19 condition patients with high anti-SARS-CoV-2 IgE antibodies had related impaired ability to perform basic activities of daily living ( Figure 5 ). This raises the possibility that a derangement of patient´s immune system is an underlying cause of, and/or that such a serological profile is reflecting long COVID patients compromised health. This study highlights an abnormal response to a virus with an IgE isotype, normally seen in parasitic infections or immunoallergic reactions, providing an additional indication that SARS-CoV-2 induces peculiar immune reactions and is likely to dysregulate both cellular and humoral immunity. In fact, it is already known that SARS-CoV-2 infection involves lymphocyte impairment leading to lymphopenia in severe acute cases, frequent hyperneutrophilia, and hyperactivation of innate-immunity (55–57). However, the antibody detection appears to decrease to the limits of detection over time as seen with the post-infection and post-vaccination delays, which may limit such serological analyses and correlations to several months after SARS-CoV-2 infection or would require a more sensitive technique to detect these antibodies on a longer period. Anyhow, long-COVID (or post-COVID) diagnostic and therapeutic intervention, when made possible on the bases of precision medicine for this rather heterogeneous nosological entity, should take place within this relatively early period and not too late or when chronicity of symptoms has been established with modifications in the pattern of biomarkers, which may then create difficulties to establish a link with previous SARS-CoV-2 infection. A limiting aspect that quite likely has impacted the study of CFS and other post-viral syndromes. The present pilot results therefore suggest that precision medicine becomes possible for such disease conditions, which future studies should elaborate. After this cross-sectional study has been performed for a pilot evaluation, studies with larger series are foreseen to include a longitudinal analysis to analyze the level of expression of both HERV-W ENV and IgE proteins and certain clinical characteristics/symptoms of long/post- COVID-19. This could however not be envisaged and set up without the rationale now provided by these first results.
The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.
The studies involving human participants were reviewed and approved by Dirección General de Salud Pública-Centro Superior de Investigación en Salud Pública (DGSP-CSISP) of Valencia, Valencia, Spain, núm. 20210604/04/01 and Centre de Resource Biologique de Hospices Civiles de Lyon, Hôpital de la Croix-Rousse, Lyon France, Autorisation N°: DC-2020-3919 and AC-2020-3918. The patients/participants provided their written informed consent to participate in this study.
KG-O: data analysis, investigation, data curation, figure drawing, writing original draft and manuscript review. JP: serological and antigenemia analyses; graphs, figures drawing and manuscript review. JB: serological and antigenemia analyses; graphs, figures drawing and manuscript review. BC: methodology, management interpretation of analyses and manuscript review. EM-M: patient diagnosis, data gathering and manuscript review. HP and EO: conceptualization, funding acquisition, supervision, formal analysis, investigation, data curation, writing original draft and manuscript review. All authors contributed to the article and approved the submitted version.
This study was funded by an Star Exclusivas SL grant, an ME Research UK (SCIO charity number SC036942) grant and by Generalitat valenciana CIAICO/2021/103 grant to EO. KG-O is supported by the Generalitat Valenciana ACIF2021/179 grant. Funders were not involved in any of the research stages.
We want to particularly acknowledge the patients and the IBSP-CV Biobank (PT17/0015/0017) integrated in the Spanish National Biobanks Network for their collaboration. Authors also want to express their gratitude to Dr. Vicente Serra (Umivale, Valencia, Spain) for his help in the recruitment of volunteers.
Authors JP, JB, BC and HP were employed by company GeNeuro-Innovation. HP, BC, JB and BC receive compensation from GeNeuro-Innovation for their work. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647064 | Yang Shi,Meiqi Wang,Dan Liu,Saif Ullah,Xing Ma,Huiyu Yang,Bingrong Liu | Super-enhancers in esophageal carcinoma: Transcriptional addictions and therapeutic strategies | 27-10-2022 | super-enhancers,esophageal carcinoma,core regulatory circuitry,transcriptional dependence,histone acetylation | The tumorigenesis of esophageal carcinoma arises from transcriptional dysregulation would become exceptionally dependent on specific regulators of gene expression, which could be preferentially attributed to the larger non-coding cis-regulatory elements, i.e. super-enhancers (SEs). SEs, large genomic regulatory entity in close genomic proximity, are underpinned by control cancer cell identity. As a consequence, the transcriptional addictions driven by SEs could offer an Achilles’ heel for molecular treatments on patients of esophageal carcinoma and other types of cancer as well. In this review, we summarize the recent findings about the oncogenic SEs upon which esophageal cancer cells depend, and discuss why SEs could be seen as the hallmark of cancer, how transcriptional dependencies driven by SEs, and what opportunities could be supplied based on this cancer-specific SEs. | Super-enhancers in esophageal carcinoma: Transcriptional addictions and therapeutic strategies
The tumorigenesis of esophageal carcinoma arises from transcriptional dysregulation would become exceptionally dependent on specific regulators of gene expression, which could be preferentially attributed to the larger non-coding cis-regulatory elements, i.e. super-enhancers (SEs). SEs, large genomic regulatory entity in close genomic proximity, are underpinned by control cancer cell identity. As a consequence, the transcriptional addictions driven by SEs could offer an Achilles’ heel for molecular treatments on patients of esophageal carcinoma and other types of cancer as well. In this review, we summarize the recent findings about the oncogenic SEs upon which esophageal cancer cells depend, and discuss why SEs could be seen as the hallmark of cancer, how transcriptional dependencies driven by SEs, and what opportunities could be supplied based on this cancer-specific SEs.
Esophageal carcinoma is the fourth most common gastrointestinal cancer worldwide. The incidence and mortality rates of esophageal carcinoma accounts for 11.75% and 14.99%, respectively (1). According to statistics from the International Researches Agency of Cancer (IRAC), the mortality rate of the cancer would increase 69% in China by 2040, while the main histological type is squamous cell carcinoma (ESCC) (2, 3). The standard-of-care regimens consisted of surgery, platinum-based chemotherapy, radiotherapy, and PD-1/PD-L1 blockade therapy (4, 5), which have improved the survival rate and patient’s quality of life. However, chemoradiotherapy resistance, postoperative recurrence and unresectable advanced lesions are still impeding the long-term survival of these patients. Given that the limited actionable drivers in ESCC, more effective avenues based on the clarification of the carcinogenesis are required. Recently, it is widely recognized that the disease-associated variations are bound up with enhancer regions, especially super-enhancers (SEs), which undoubtedly provides novel insights into therapeutic maneuvers. Enhancers refer to non-coding part of genome that activate genes expression independent of orientation, distance, and location regarding its transcription start sites (TSSs). These regulatory elements always provide binding sites for multiple transcription factors (TFs) that can also be transcribed to produce non-coding enhancer RNAs (eRNAs). More recently, cluster of enhancers located genomic proximity were identified as the unique transcriptional single entity, known as SEs. SEs are underpinned by highly abundant and orchestrated interactions with transcription apparatus and active enhancer marks (Acetylation at lysine 27, H3K27ac & Monomethylation at lysine 4, H3K4me1). Notably, SEs have been demonstrated to dictate cell identity and disease and play a key role in carcinogenesis in a broad spectrum of tumors. Recently, it has been proposed that SEs have the potential to serve as valuable prognostic and therapeutic targets in cancer. During the past decade, various types of cancer pathogenesis have been proved to be closely associated with SEs, such as oncogenes activation, dysregulated signaling pathways, and genetic mutations. Generally, cancer-specific SEs assembly are not presented in the corresponding non-cancerous tissues, such as C-Myc, INSM1 (6) and TAL1 (7). Broadly, these SEs activate the tumorigenic signaling pathways, promoting oncogenic transcriptions, and enriched in key TFs binding motifs. Fortunately, the sensitivity of SEs to perturbation has shown the promising therapeutic vulnerability in various types of cancer, including breast cancer (8), nasopharyngeal carcinoma (9), small cell lung cancer (10), medulloblastoma (11) and esophageal carcinoma (12). The critical oncogenes of ESCC, e.g. TP63, SOX2, KLF5 and ALDH3A1, have been shown to participate in core regulatory circuitry (CRC) driven by SEs (12). Besides, the pharmacological inhibition of cyclin dependent kinase 7 (CDK7), bromodomain-containing protein 4 (BRD4) and histone deacetylases (HDACs), has been applied in esophageal carcinoma treatment (13). Therefore, not surprisingly, deregulation of SEs is fundamental mechanism of cancer, which offers an Achilles heel for diagnostic and therapeutic maneuvers (14). This review attempts to discuss SEs’ fundamental characteristics and roles in esophageal carcinoma, which would pave the way for SE-based diagnostic and therapeutic maneuvers.
Enhancers were firstly recognized as the cis-regulatory elements from simian virus 40 (SV40), which could prominently promote rabbit β-globulin transcription in HeLa cells (15, 16). In general, enhancers activate cell-type-specific gene expression regardless of their distance, position, and orientation with respective to the cognate promoter (17, 18). The elements of enhancer are bound up with critical TFs through their tissue-specific recognition motifs, thereby functioning as the platform to integrate signaling pathways and further dictate cell lineages (19). Mechanistically, the enrichment of master TFs on enhancers results in recruiting of the subunit of the Mediator complex (Med1), RNA polymerase II (Pol II), and the basal transcription apparatus, which is organized by looping between enhancers and their cognate promoters. Additionally, enhancer regions are mainly overlapped with DNase hypersensitive sites (DHS), and the active state of enhancers are dependent on the following combinations of histone modifications: enrichment with H3K27ac, H3K4me1, and deficiency of Trimethylation at lysine 4 (H3K4me3) ( Table 1 ) (20). SEs were firstly identified as the unique cluster of enhancers in close genomic proximity, which were densely occupied by master TFs Oct4/Pou5f1, Sox2, Nanog (OSN), Klf4, Esrrb, and Med1 in murine embryonic stem cells (mESCs) (21, 22). The Rank Ordering of Super-Enhancers (ROSE) algorithm had been proposed to separate typical-enhancers (TEs) from SEs, whose constituent enhancers were stitched together within 12.5 kb genomic regions enriched by input-normalized level of Med1 signal.
Liquid-liquid phase separation (LLPS) is a physicochemical process by which membraneless organelles are generated in eukaryotic cells, which could compartmentalize biochemical reactions within the dense phase (23–25). It has been demonstrated that SEs are phase-separated assemblies accumulated by exceptionally high densities of master TFs, co-activators, and RNA Pol II, dictating the roles in cell identity and disease, including cancers (26). The intrinsically disordered regions (IDRs) of the co-activators (BRD4 & Med1), driven by high-valency and weak-affinity interactions, are responsible for the formation of the phase-separated condensates at SEs, which could bring those cis-elements and the cognate promoter in close three-dimensional (3D) proximity and then facilitate SEs activation ( Figure 1 ). This physical interaction between enhancer and promoter both involving in TEs and SEs are mediated by cohesin-associated CTCF (CCCTC-binding factor) loops. Additionally, the ubiquitously expressed TF Yin Yang 1 (YY1) has been also identified as the structural regulators contributing to this loop structure (27). The disruption of this structure is likely to result in activation of oncogenes outside the neighborhood by deletion of CTCF binding site, which are consistent with a tendency of liquid phase condensates to undergo fusion.
The main features of SEs had been summarized as following Richard A. Young et al. and colleagues: (i) high-density occupancy of master TFs, co-activators and chromatin remodelers ( Table 1 ), (ii) large genomic spanning, (iii) ability to exceptionally activate transcription, (iv) sensitivity to perturbation and (v) dictate cell identity and disease (28). Based on these observations, the oncogenic role of SEs usually caused by genetic variations are proposed to drive the transcriptional addiction in cancer. The transcriptional activities of SEs exhibit an order-of-magnitude higher than TEs and the individual constituent enhancers within SEs, which also showing highly context-dependent manner under rigorous genetic regulation (29). And more interestingly, the cooperativity of SEs constituents is neither additively nor synergistically, suggesting that the “modus operandi” of each component under highly precise and complex regulation by other component (30).
Generally, inappropriate SEs are acquired de novo during tumorigenesis compared with the counterpart normal tissues, which driving expression of the critical oncogenes. The malignant transformation and maintenance underpinned by tumorigenic SEs could be seen as one of the core tenets of cancer biology (28, 31, 32). Mechanically, the formation of oncogenic SEs may stem from (i) focal amplification, (ii) genomic rearrangements, (iii) single nucleotide polymorphisms (SNPs), (iv) disruption of topological associating domain (TADs).
Somatic copy number alterations (SCNAs) are common mechanism of oncogenesis driven by SEs. For example, focal amplification peak harboring SEs identified by profiling of H3K27ac in esophageal carcinoma (chr13:73880413-74042621, about ~162 kb), which subsequently proven to activate the oncogene KLF5 expression. And these cancer-specific focal amplification peaks enriched in SEs have also been identified in several types of cancers, including head and neck squamous cell carcinoma, colorectal carcinoma, and liver hepatocellular carcinoma as well (33). Additionally, genomic rearrangements could change the natural genomic context resulting in close proximity between oncogene promoter and their SEs, which ectopically activates gene expression (34). This phenomenon is also described as “enhancer hijacking”, by which oncogenic SEs formed in colorectal carcinoma, medulloblastoma, and acute myeloid leukemia as well (11, 35, 36). Besides, SNPs have been reported to promote tumorigenesis by disrupting the activities of SEs, including acquired oncogenic or abrogating protective allele within master TFs binding sites (37, 38). These genomic variants provide novel insights into the carcinogenesis of esophageal carcinoma, which deserves further investigation.
TADs have been recognized as the self-interacting genomic region, which demonstrates higher interconnection frequency than the outer regions (39). It has become clear that the main function of TADs is to insulate promoters from distal enhancers or SEs, which conducted by binding of insulator CCCTC-binding factor (CTCF) in cooperation with cohesin complex at their TAD boundaries (40). Although TADs structures are conserved in mammalian, disruption of TADs boundaries caused by genetic or epigenetic could be convenient for abnormal interactions between enhancer/SEs and promoters, which is undoubtedly could be laid the foundation for tumorigenesis. It has been reported that TADs boundaries disruption could be caused by recurrent mutations of CTCF and cohesin binding sites, which were identified in esophageal carcinoma as well as other types of cancers, including liver hepatocellular carcinoma, colorectal carcinoma, and gastric carcinoma (41, 42). Furthermore, the boundaries disruption is also exemplified by epigenetic dysregulation in glioma by increasing hypermethylation in CTCF site followed by its reduced binding activity (43). Modification of 3D genome by TADs disruption could activate oncogene expression driven by inadvertent SEs-promoter looping, which might be utilized as the novel candidates for targeting the oncogenic SEs in esophageal carcinoma.
The ESC master TFs OSN have previously been demonstrated to bind to their own genes or those of the others in mESCs, which forms an autoregulatory feed-froward loop, i.e., CRC (44, 45). The constituents of this interconnected loop were subsequently extended with the other core TFs, Klf4 and Esrrb, both of which prominent for the maintenance of ESC state ( Figure 2A ) (21). Thus, CRC plays a critical role in the reprogram of somatic cell into induced pluripotent stem cells (iPSCs), and its dysregulation is undoubtedly involved in cancer (12, 46). Trio-occupancy of OSN has also been well-studied in hESCs (human ESCs), which dominates the pluripotency contributing to human fibroblasts differentiated into the induced pluripotent identity. And this model has been complemented for additional seven key TFs by “CRC Mapper” algorithm, including FOXO1, ZIC3, NR5A1, RARG, MYB, RORA, and SOX21 ( Figure 2B ) (47). The principle of CRC identification by “CRC Mapper” has been characterized as the following three properties: TFs (i) encoded by SEs-assigned genes, (ii) binding to SEs of their own genes, (iii) forming fully interconnected feed-froward loops with other TFs by binding to their SEs. In accordance with this strategy, cancer-type and -subtype specificity of CRC models have been identified in esophageal cancer and other malignancies. For example, KLF5 has been proven to be collaborated with ELF3, GATA6 and EHF in EAC (46), while cooperated with TP63 and SOX2 in ESCC to form CRC (12), respectively ( Figure 2C ). And SOX2 has also been identified co-regulated with KLF4, EGR1 and NOTCH1 in Glioblastoma (48). These cancer-type and -subtype specific CRCs driven by SEs orchestrate the oncogenic transcriptional addiction, while offers therapeutic vulnerabilities due to perturbation sensitivity of their own SEs. These tissue-specific and cancer-specific CRCs driven by SEs orchestrate the oncogenic transcriptional addiction, which offers therapeutic vulnerabilities consistent with the perturbation sensitivity of their own SEs. Furthermore, cell-type-specific CRC observed in types of cancers is in line with tumor heterogeneity, one of the hallmarks of malignancy. For example, the other CRC (MYC/JUNB/FOSL1) has been identified in ESCC, different with the KLF5/SOX2/TP63 circuitry ( Figure 2C ). This heterogeneity of CRC coincides with the two major genomic molecular subtypes in ESCC, which are presumably dominated by the two CRCs respectively. And different CRCs have also been identified in other types of cancers, including medulloblastomas and acute myeloid leukemia (49, 50). The heterogeneity of CRC highlights cell-type-specific property of SEs, which sheds light on the novel therapeutic possibilities.
The constituents within SEs have been demonstrated to be heavily loaded with terminal TFs of the Wnt, Tgfb-1, and Lif signaling pathways in mESCs, showing the preferential affections upon SE-assigned genes compared with genes regulated by TEs (51). Therefore, SEs could serve as a platform to converging multiple signaling pathways, dictating the development and disease state of cells, especially oncogenesis (52). For example, EAC-specific SEs has been proposed to be loaded with tumor-promoting TFs LIF, which contributes its malignant features by activating STAT3 and PI3K/AKT signaling pathways (46). Similarly, the cancer-subtype-specific SEs are densely bound up with its specific master TFs TP63 in ESCC, which participates in the formation of CRC and promotes cancer cell proliferation via PI3K/AKT signaling pathways (12, 53, 54). These findings have also been clarified in other types of cancer, e.g., colorectal carcinoma (51), pancreatic carcinoma (55) and osteosarcoma (56). Taken together, these lines of evidence have been proposed that these oncogenic SEs occupied by key TFs pertaining to the critical signaling pathways upon which cancer cell depend, which offers the therapeutic vulnerabilities for esophageal carcinoma.
The cyclin-dependent kinases (CDKs) of mammals comprise two main subfamilies with unique properties associated with cell cycle (CDKs 1-6 & 14-18) and transcriptional regulation (CDKs 7-13 & 19-20) (57). The unique functional repertoire of CDK7, the critical component of CDKs, is based on the regulation of transcription and cell cycle progression. Generally, CDK7 could activate transcriptional initiation and elongation by combining with transcription factor II H (TFIIH), while it could also control cell cycle progression by virtue of forming CDK-activating kinase (CAK) (58, 59). Studies have demonstrated that the dysregulation of CDK7 was involved in various types of cancers and considered to be positively correlated with the aggressive clinicopathological features of these cancers, including esophageal cancer, hepatocellular carcinoma, gastric cancer, and colorectal cancer (13, 60–62). In ESCC, an immunohistochemical (IHC) analysis demonstrated that elevated expression of CDK7 was observed in over 80% samples which was associated with high tumor grade and poor prognosis. Notably, inhibited proliferation and decreased chemotherapeutic resistance have been observed when CDK7 gene was silenced (63). Besides, the level of CDK7 was higher in ESCC tissues with lymph node metastases compared to control group and positively correlated with tumor metastasis and patients’ overall survival (64). The overabundance of CDK7 within SEs regions offer the opportunities of blockade therapies in a lineage-specific cancer cell manner. THZ1, a covalent inhibitor of CDK7, was proposed to have a preferential impact on a plethora of oncogenes driven by SEs and could cause the disruption of specific transcriptional programs. For example, Chipumuro et al. found that this inhibitor could suppress the transcription of amplified Mycn to suppress the neuroblastoma cells proliferation and the sensitivity to THZ1 was related to preferentially decreased expression of SE-driven oncogenes (65). Subsequent studied showed that esophageal cancer, lung cancer and prostate cancer were sensitive to THZ1 treatment at low nanomolar range (10, 13, 66). Additionally, THZ1 could also cause downregulation of SEs-associated functional long noncoding RNAs acting as competing endogenous (ce-lncRNAs), such as HOTAIR, XIST, SNHG5, and LINC00094 (67), which are associated with expression of cancer hallmark genes. Notably, LINC00094, as a novel oncogenic lncRNA could be activated by master TFs, e.g., TCF3 and KLF5, and positively correlate with poor prognosis. Upon inhibiting these TFs by THZ1, the level of LINC00094 is downregulated, which could cause tumor regression in ESCC. Thus, these studies have demonstrated THZ1 could be utilized as the crosshair of cancer drug discovery (57). Similar to CDK7, CDK9 is mainly responsible for oncogenic transcription, e.g., as MCL-1 and C-Myc, by binding to elongation complex (p-TEFb) (68) and knocking down CDK9 has shown an excellent antineoplastic activity in hematologic and solid tumors. SNS-032, a selective inhibitor against CDK family as well can inhibit transcription initiation and elongation by targeting both CDK7 and CKD9, with the IC50 values of 62 and 4 nM, respectively (69–71). It could suppress the lung and lymph node metastasis as well as inhibit ESCC proliferation (72). These findings indicate that SNS-032 play an antineoplastic role in ESCC and is expected to enter clinical trials to validate its efficacy, especially in those with metastasis.
The bromo domain and extra-terminal domain (BET) family contain BRD2, BRD3, BRD4, and testis-specific BRDT, which are characterized by acetylation recognition and transcription regulation (73). The well-known member of the BET family, BRD4 was initially recognized as the component of Mediator complex, and subsequently the general regulator for RNA Pol II by virtue of recruitment with transcriptional elongation factor P-TFEb. Consistently, the genomic profiling of BRD4 demonstrated that it is mainly enriched at active enhancers and promoters, which is not surprisingly complicated in critical cellular processes, such as embryogenesis (74) and cancer development (75–82). More importantly, the tumorigenic transcriptional activation of BRD4 are preferentially to regulate SE associated oncogenes, such as C-Myc and BCL2 (83, 84) Emerging evidence have demonstrated that inhibition of BRD4 occupancy result in SEs disruption and subsequently suppress its related oncogenes expression (73). Up till now, about 20 BRD4 inhibitors have been entered into clinical trials, some of which showed valuable therapeutics for several cancers, including hematological malignancies and non-small cell lung cancer (85). The critical oncogenes within these cancers showed highly sensitivity to JQ1, one of the promising anti-tumor BET inhibitors (22, 86). For instance, JQ1 has been demonstrated to inhibit ESCC proliferation by inhibiting C-Myc amplification (86). For another, it could also block recruitment of BRD4 on the promoter of aurora kinases A and B (AURKA/B) to trigger cellular senescence, which provided a novel action manner of BRD4 in esophageal carcinoma (87). In esophageal adenocarcinoma (EAC), high expression of YAP1 has been reported to be observed and positively associated with poor prognosis. JQ1 is capable of blocking BRD4 binding to the YAP1 promoter and suppress Hippo/YAP1 signaling (88), which could be synergistically enhanced when traditional chemotherapeutic agents, e.g., docetaxel, are added both in vitro and in vivo (88). Notably, cell cycle arrest, especially in G1 phase, has been observed in all the studies above. These studies also indicated that JQ1 might be a promising drug in esophageal carcinoma treatment. Although a variety of small-molecule BET inhibitors have been entered in clinical researches, however, the off-target and side-effects cannot be neglected (89). Proteolysis targeting chimeric (PROTAC) technology have been utilized as an effective degradation tool over the years, which can ubiquitinate the disease-causing proteins by hijacking E3 ligases to achieve the anti-tumor effects (90). Compared to the general small-molecule inhibitors, the hetero-bifunctional molecule based on PROTAC technology demonstrated degradation of BET protein at low nanomolar dose with negligible side-effects (91). The first oral PROTACs (ARV-110, NCT03888612 & ARV-471, NCT04072952) have achieved encouraging benefit for prostate and breast cancer treatment in clinical trials (92). The newly developed GNE987, and PROTAC pan-BET degrader, have been showing good potency against several cancers, including hematological malignancies (93, 94), neuroblastoma (95), and prostate carcinoma (96). Additionally, MZ1, a PROTAC BET degrader, was previously identified unexpectedly degradation of BRD4 over other members of BET family, such as BRD2 and BRD3 (97). However, MZ1 have been proposed to suppress ESCC migration by degrading BRDT, initially recognized as testis-specific protein, rather than BRD2, BRD3 and BRD4. The migratory inhibition of ESCC by MZ1 resulted from down-regulation of ΔNp63 target genes enriched within SEs (98).
Histone H3K27ac has been proposed to separate the active enhancers from poised enhancers occupied by H3K4me1 alone (99). The occupancy of H3K27ac facilitates the chromatin accessibility, which is responsible for exceptionally higher enrichment of master TFs, co-activators, and transcriptional machinery on SEs. Therefore, the dysregulated H3K27ac within SEs region could drive the oncogenesis in several types of cancer. Histone acetylation is a relative steady-state controlled by two families of enzymes: histone acetyltransferases (HATs) and histone deacetylases (HDACs) (100, 101). Numerous studies have demonstrated that the disturbed equilibrium of histones acetylation is closely related with the tumorigenicity of esophageal carcinoma ( Table 2 ). Furthermore, the tumor-promoting genes activated by hyperacetylation have also been identified in esophageal carcinoma ( Table 3 ). The dysregulated histone acetylation of esophageal carcinoma could be accounted for the oncogenesis driven by SEs, which densely bound up with the active enhancer marker H3K27ac. Based on this knowledge, the aberrant histone acetylation has been addressed as the alternative avenues for cancer treatment, especially those driven by oncogenic SEs. For example, the aberrantly expressed gene SIRT7, NAD-dependent deacetylase, was activated by SEs in non-alcoholic fatty liver disease (NAFLD)-associated hepatocellular carcinoma (HCCs). Depletion of SIRT7 associated SE could suppress the tumorigenicity both in vitro and in vivo (110). And several types of HDAC inhibitors (HDACis) have been shown as the promising therapeutic strategies against esophageal carcinoma ( Table 4 ). Therefore, the recruitment of exceptionally high histone acetylation modification, especially H3K27ac, within the oncogenic SEs could contribute their vulnerabilities to HDACis for patients of esophageal carcinoma.
The concept of SEs and their function in physiology and disease has been established in the last decade. To date, there is no doubt that SEs acquired de novo are one of the hallmarks of esophageal carcinoma, and the available studies conclude the characteristics of the oncogenic SEs in esophageal carcinoma: (i) they are acquired based on genetic variants or TADs boundaries disruption, (ii) they drive the transcriptional additions depend on CRCs and convergence of tumorigenic signaling pathways, (iii) they could be targeted by inhibition of BRD4, CDK7, and histone acetylation, which exceptionally enriched within these cis-elements. However, there are still several unresolved points regarding SEs in esophageal carcinoma: (i) how the individual bona fide constituents selectively respond to targeted inhibition pertaining to the disproportionately enriched factors, (ii) how 3D genome structure of the SEs are influenced by the tumor-promoting signaling pathways, (iii) what are the critical transitions of SEs landscape during the process of precancerous lesions to cancer, (iv) what are the main differences of SEs profiling between squamous cell carcinoma and adenocarcinoma, (v) what are the clinical benefits based on the combinations of traditional treatments and novel therapeutic avenues targeting the subtype-specific or cell-type-specific CRCs. These insights about SEs-driven transcriptional dependencies in esophageal carcinoma, may shed light on the potential clinical application of the cancer-specific SEs.
BL, HY, and XM formulated the study concept and design; YS wrote the original manuscript draft; MW, DL, provided the technical and material support; SU, critically revised the manuscript. All authors contributed to the article and approved the submitted version.
This study was supported by the Zhongyuan talent program (ZYYCYU202012113).
We apologize to the authors whose study could not be cited for the limited spaces. We would like to thank Jiyu Zhang for his delicate and exquisite schematic graphs. And we also thank Zhixin Li for his comments on this manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647074 | 35618276 | Yu Sun,Mei Zhang,Zheyuan Ou,Yi Meng,Yang Chen,Ruqin Lin,Jamal Hisham Hashim,Zailina Hashim,Gunilla Wieslander,Qingsong Chen,Dan Norbäck,Xi Fu | Indoor microbiome, microbial and plant metabolites, chemical compounds, and asthma symptoms in junior high school students: a multicentre association study in Malaysia | 10-11-2022 | Background Indoor microbial exposure is associated with asthma, but the health effects of indoor metabolites and chemicals have not been comprehensively assessed. Methods We collected classroom dust from 24 junior high schools in three geographically distanced areas in Malaysia (Johor Bahru, Terengganu and Penang), and conducted culture-independent high-throughput microbiome and untargeted metabolomics/chemical profiling. Results 1290 students were surveyed for asthma symptoms (wheeze). In each centre, we found significant variation in the prevalence of wheeze among schools, which could be explained by personal characteristics and air pollutants. Large-scale microbial variations were observed between the three centres; the potential protective bacteria were mainly from phyla Actinobacteria in Johor Bahru, Cyanobacteria in Terengganu and Proteobacteria in Penang. In total, 2633 metabolites and chemicals were characterised. Many metabolites were enriched in low-wheeze schools, including plant secondary metabolites flavonoids/isoflavonoids (isoliquiritigenin, formononetin, astragalin), indole and derivatives (indole, serotonin, 1H-indole-3-carboxaldehyde), and others (biotin, chavicol). A neural network analysis showed that the indole derivatives were co-occurring with the potential protective microbial taxa, including Actinomycetospora, Fischerella and Truepera, suggesting these microorganisms may pose health effects by releasing indole metabolites. A few synthetic chemicals were enriched in high-wheeze schools, including pesticides (2(3H)-benzothiazolethione), fragrances (2-aminobenzoic acid, isovaleric acid), detergents and plastics (phthalic acid), and industrial materials (4,4-sulfonyldiphenol). Conclusions This is the first association study between high-throughput indoor chemical profiling and asthma symptoms. The consistent results from the three centres indicate that indoor metabolites/chemicals could be a better indicator than the indoor microbiome for environmental and health assessments, providing new insights for asthma prediction, prevention and control. | Indoor microbiome, microbial and plant metabolites, chemical compounds, and asthma symptoms in junior high school students: a multicentre association study in Malaysia
Indoor microbial exposure is associated with asthma, but the health effects of indoor metabolites and chemicals have not been comprehensively assessed.
We collected classroom dust from 24 junior high schools in three geographically distanced areas in Malaysia (Johor Bahru, Terengganu and Penang), and conducted culture-independent high-throughput microbiome and untargeted metabolomics/chemical profiling.
1290 students were surveyed for asthma symptoms (wheeze). In each centre, we found significant variation in the prevalence of wheeze among schools, which could be explained by personal characteristics and air pollutants. Large-scale microbial variations were observed between the three centres; the potential protective bacteria were mainly from phyla Actinobacteria in Johor Bahru, Cyanobacteria in Terengganu and Proteobacteria in Penang. In total, 2633 metabolites and chemicals were characterised. Many metabolites were enriched in low-wheeze schools, including plant secondary metabolites flavonoids/isoflavonoids (isoliquiritigenin, formononetin, astragalin), indole and derivatives (indole, serotonin, 1H-indole-3-carboxaldehyde), and others (biotin, chavicol). A neural network analysis showed that the indole derivatives were co-occurring with the potential protective microbial taxa, including Actinomycetospora, Fischerella and Truepera, suggesting these microorganisms may pose health effects by releasing indole metabolites. A few synthetic chemicals were enriched in high-wheeze schools, including pesticides (2(3H)-benzothiazolethione), fragrances (2-aminobenzoic acid, isovaleric acid), detergents and plastics (phthalic acid), and industrial materials (4,4-sulfonyldiphenol).
This is the first association study between high-throughput indoor chemical profiling and asthma symptoms. The consistent results from the three centres indicate that indoor metabolites/chemicals could be a better indicator than the indoor microbiome for environmental and health assessments, providing new insights for asthma prediction, prevention and control.
Asthma is a common allergy-related chronic respiratory disease that affects more than 350 million patients worldwide. The prevalence of asthma symptoms (wheeze and whistling) is >30% in many countries, including Australia, Ireland and the UK, posing a severe threat to public health [1]. Epidemiology studies show that the occurrence of asthma and allergies is mainly affected by environmental exposure, including air pollution, environmental allergens and microorganisms [2–4]. A striking phenomenon is that the prevalence of asthma is significantly lower in children growing up in farming or rural areas than in urban areas [5]. Subsequent studies revealed that indoor microbial exposure is the driving factor for the variation [6]. Studies in Finland, Germany, Malaysia and China also confirmed the importance of the indoor microbiome in immune modulation and disease development [7–10]. However, it is challenging to transform the theoretical progress into practical applications, such as building an indoor microbiome indicator for environmental assessment and disease prediction. This is because the diversity of the indoor microbiome is extremely high. The total number of microbial species on Earth is approximately 1 trillion [11]. Also, the indoor microbiome shows extremely high geographic diversity. Different sets of indoor microorganisms and health-associated microorganisms are characterised in different geographical regions [7–10]. It is almost impossible to find a consensus set of health-related species across the globe and make a solid indoor microbiome–health inference. Thus, an alternative environmental assessment indicator is needed. Indoor metabolites and chemicals could be a potential alternative for environmental assessment. Each bacterial and fungal organism can release thousands of metabolites into the living environment per hour, affecting the health of occupants. The common health-related metabolites include lipopolysaccharide (LPS), muramic acid and microbial volatile organic compounds (MVOCs) [12, 13]. However, previous studies used culture-dependent or low-throughput approaches to characterise a small set of targeted chemical exposures from microorganisms. No study used a high-throughput untargeted approach to profile comprehensive indoor metabolites and chemicals. Thus, the overall picture of metabolites/chemicals in the indoor environment is still unclear. Also, no study has conducted multi-omic analysis between indoor microorganisms and metabolites to identify the potential microbial sources of metabolites. In this study, we surveyed 1290 junior high school students in Johor Bahru, Terengganu and Penang in Malaysia for asthma symptoms, and collected classroom dust for culture-independent high-throughput microbiome and untargeted chemical profiling. We aimed to characterise indoor metabolic/chemical exposures and uncover their relationships with health-related microorganisms. Also, we compared the environmental chemical pattern in multiple centres and tested whether it could be a better indicator than the indoor microbiome for exposure assessment.
We conducted classroom dust sampling and health surveys in three areas in Malaysia: Johor Bahru, Terengganu and Penang. The locations are displayed in supplementary figure S1. In each centre, eight junior high schools (four classrooms in each school) were randomly selected for dust sampling. In each class, health questionnaires in Malay were sent to 15–20 students aged 14–15 years. Health questions were obtained from the International Study of Asthma and Allergies in Childhood study [14], including an asthma symptom question: “In the last 12 months, have you had wheezing or whistling in the chest when you DID NOT have a cold or the flu?”. Personal information was collected, including age and gender. The participants had no information regarding environmental data and samples collected. The study design and protocol were approved by the Medical Research and Ethics Committee of the National University of Malaysia (Selangor, Malaysia). Informed consent was obtained from all participants.
We sampled classroom dust on floors, desks, chairs, bookshelves and curtains with a vacuum cleaner in Johor Bahru and Terengganu [7, 15]. The vacuum procedure was maintained at 4 min: 2 min on the floor and 2 min on other surfaces above floor level, including student desks, chairs, bookshelves and curtains. The dust was collected in a sampler (ALK-Abelló, Copenhagen, Denmark) with a filter pore size of 6 µm. In schools in Penang, we collected settled dust on the upper frame of the blackboard with a metal spoon. The vacuumed and settled dust was sieved to fine dust through a metal mesh screen (pore size 0.3 mm). The fine dust was stored in a freezer at −80°C. Indoor relative humidity and carbon dioxide were measured by a Q-Trak indoor air quality monitor (TSI, St Paul, MN, USA). Indoor nitrogen dioxide was sampled by a diffusion sampler from IVL Swedish Environmental Research (Gothenburg, Sweden).
Bacterial and fungal amplicon sequencing was performed by using dust samples. In brief, total microbial DNA was extracted from 10 mg fine dust by an E.Z.N.A. Soil DNA Kit (D5625-01; Omega Bio-Tek, Norcross, GA, USA) and a DNA SPIN Kit (MP Biomedicals, Santa Ana, CA, USA). DNA quality was assessed by a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and agarose gel electrophoresis. Bacterial 16S rRNA genes and fungal internal transcribed spacer regions were amplified, and sample-specific barcode sequences were added during the library preparation step. The amplicons were sequenced by Illumina MiSeq and PacBio platforms. Raw sequence data were deposited in the QIITA microbial study management platform (https://qiita.ucsd.edu; 12875) and the Genome Sequence Archive (https://ngdc.cncb.ac.cn/gsa; CRA002825, CRA002876, CRA005646 and CRA005647) [16]. The absolute bacterial and fungal concentration was quantified by quantitative PCR with universal primers [15].
Chemical compounds in classroom dust were assessed by untargeted liquid chromatography-mass spectrometry (LC-MS) (BioNovoGene, Suzhou, China). 10 mg fine dust was added to 0.6 mL 2-chlorophenylalanine in methanol and centrifuged at 12 000 rpm at 4°C for 10 min. 300 μL supernatant was filtered through a 0.22 μm membrane. Chromatographic separation was performed by Acquity UPLC HSS T3 columns (2.1×150 mm, 1.8 μm; Waters, Milford, MA, USA) at 40°C at a flow rate of 0.25 mL·min−1. A Vanquish HPLC systems Q Exactive HF-X hybrid quadrupole–orbitrap mass spectrometer was used for LC-MS detection (Thermo Fisher Scientific). Electrospray ionisation MS experiments were performed with spray voltages of 3.5 kV and −2.5 kV in positive and negative modes. Sheath and auxiliary gas and capillary temperature were set at 30 and 10 arbitrary units and 325°C. The analyser scanned a mass range of m/z 81–1000 at a mass resolution of 60 000. Chemicals were annotated by searching against mzCloud (www.mzcloud.org), Human Metabolome Database (www.hmdb.ca), MoNA (https://mona.fiehnlab.ucdavis.edu), METLIN (https://metlin.scripps.edu) and MassBank (www.massbank.jp).
Microbiome data processing and analysis were conducted on the QIIME 2 platform [17]. Raw reads were assigned to samples according to the barcode information. Low-quality and chimeric reads were removed. Sequence taxonomy was annotated by the Silva (release 115) and UNITE (release 5) databases [18, 19]. LEfSe (linear discriminant analysis (LDA) effect size) analysis was conducted to characterise the enriched microbial taxa in different groups (LDA >2) [20]. Mmvec was used to estimate microbe–metabolite interaction co-occurrence probabilities [21] and the results were visualised by TBtools [22]. The key features of the chemical compounds and Globally Harmonized System (GHS) classification were obtained from the PubChem database [23].
In total, 1290 junior high school students from Johor Bahru, Terengganu and Penang were surveyed for asthma symptoms/wheeze. The three centres were located in the south, northeast and northwest Malaysia (supplementary figure S1), and eight junior high schools were randomly selected in each centre. We found large-scale variation in the prevalence of asthma symptoms among schools in Johor Bahru, Terengganu and Penang (p=0.0007, p=0.0004 and p=0.002, respectively) (table 1). For example, the prevalence of wheeze was 12.7%, 20.0%, 14.3%, 29.9%, 8.9%, 8.7%, 5.6% and 9.1% in the eight schools in Terengganu. In each centre, the top four schools were defined as “high-wheeze” schools and the bottom four schools were defined as “low-wheeze” schools. We further explored the personal and environmental characteristics that could explain the variation of asthma symptoms among schools. Students’ age and gender did not differ between high- and low-wheeze schools (p>0.05, Mann–Whitney test). The indoor environmental characteristics, including temperature, relative humidity and carbon dioxide concentration, were similar among the three centres (table 2). Indoor nitrogen dioxide concentration was lower in Terengganu than in Johor Bahru and Penang. In all centres, these environmental characteristics did not differ between high- and low-wheeze schools (p>0.05), failing to explain the variation of asthma symptoms.
High-throughput sequencing was applied to characterise the abundance of indoor microorganisms in Johor Bahru, Terengganu and Penang (supplementary tables S1–S8). We found strong geographic variation in indoor bacteria and fungi among centres. For example, bacterial genera Ralstonia, Ochrobactrum and Craurococcus were present in high abundance (7.8%, 2.7% and 1.3%, respectively) in Penang, but were presented in low abundance in Johor Bahru and Terengganu (0% and 0.06%, 0.02% and 0.42%, and 0.05% and 0.74%, respectively) (supplementary table S7). Fungal genera Wallemia and Candida were present in high abundance in Johor Bahru and Terengganu (6.6% and 12.2%, and 1.3% and 1.8%, respectively) but in low abundance in Penang (0.06% and 0.03%, respectively) (supplementary table S8). Common mould species Penicillium and Cladosporium were present in high abundance only in Johor Bahru (10.2% and 7.8%, respectively). Fusarium was present in high abundance in Terengganu but not in Johor Bahru and Penang (2.9% versus 0.3% and 0.009%). Aspergillus was present in high abundance in all centres (20.7%, 12.7% and 12.1%, respectively). We further explored the potential indoor microorganisms enriched in high/low-wheeze schools in each centre. In Johor Bahru, 15 bacterial taxa were enriched in low-wheeze schools and more than half were from phylum Actinobacteria, including Rubrobacter, Actinomyces, Blastococcus, Janibacter, Actinomycetospora, Pseudokineococcus and Marmoricola (figure 1 and supplementary figure S2). However, in Terengganu, the bacteria enriched in low-wheeze schools were mainly from phylum Cyanobacteria, including Chroococcidiopsis, Fischerella, Mastigocoleus and Iphinoe. In Penang, the bacteria enriched in low-wheeze schools were mainly from phylum Proteobacteria, including Caulobacter, Bosea, Acidovorax, Undibacterium and uncharacterised (uc)_Sphingomonadaceae. The results indicate that different centres have a unique set of potential protective bacterial taxa. The bacterial taxa enriched in high-wheeze schools also differed among centres. Catellicoccus and Ignatzschineria were enriched in high-wheeze schools in Johor Bahru, Methylobacterium, Bacillus, Sphingomonas and Pantoea were enriched in Terengganu, and Roseomonas was enriched in Penang (supplementary figure S2). The potential protective fungal taxa were from phyla Ascomycota and Basidiomycota (figure 1). Quantitative PCR was also conducted in Johor Bahru and Terengganu (supplementary tables S9 and S10). The absolute concentration of indoor bacteria and fungi did not differ between high- and low-wheeze schools (p>0.1, t-test).
In total, 2633 chemicals were characterised by LC-MS. The geographic pattern was also observed for chemical compounds (supplementary figure S3). Indoor chemical composition in Johor Bahru, Terengganu and Penang is located, respectively, in the right, upper-left and lower-left side of the ordination plot (supplementary figure S3b). However, general rules were also observed. Three classes of metabolites were almost exclusively enriched in low-wheeze schools (p<0.01, false discovery rate (FDR) <0.1, fold change >2), including flavonoids, isoflavonoids, and indole and derivatives (table 3 and figure 1). Flavonoids and isoflavonoids are important classes of plant secondary metabolites, widely found in various plants, fruits and vegetables. Three flavonoids and isoflavonoids were enriched in two or more centres, including isoliquiritigenin, formononetin and 6-hydroxydaidzein. Other flavonoid metabolites, including baicalin, astragalin, tangeritin, daidzein, luteolin and procyanidin B2, were enriched in one centre (table 3). Indole and derivatives form a class of common small signalling molecules in microorganisms, plants and animals. Serotonin and indole-3-carboxaldehyde were enriched in two or more centres, and indole, l-tryptophan, 1H-indole-3-acetamide and indolepyruvate were enriched in one centre. The explicit enrichment of flavonoids and indole derivatives in the dust of low-wheeze schools suggests their potential anti-inflammatory and antiallergic effects. A neural network analysis showed that the indole and its derivatives were co-occurring with many potential protective microbial taxa, such as l-tryptophan with Actinomycetospora and uc_Corynebacteriacea, N-acetylserotonin and indole with Truepera, indole with uc_Pleosporaceae, and 5-methoxyindoleacetate with Fischerella. The results suggest that these microorganisms may produce these metabolites. A literature search showed that Actinomycetospora and Fischerella were capable of producing indole derivatives [24, 25], supporting the in silico association results. Other potential protective metabolites were also identified, including biotin, chavicol, ecgonine, dihydrocortisol, etc. (figure 2), which belong to different metabolic classes. Biotin was closely associated with many protective bacteria, including Actinomyces, Paracoccus and Sphingomonas. A literature search in laboratory experiments showed that these taxa could produce biotin [26–28], consistent with the co-occurrence analysis.
Potential risk environmental chemicals were defined as chemicals significantly enriched in high-wheeze schools (p<0.01, FDR <0.1, fold change >2). One (2(3H)-benzothiazolethione), one (2-aminobenzoic acid) and three (isovaleric acid, phthalic acid and 4,4-sulfonyldiphenol) hazardous chemicals were characterised in Johor Bahru, Terengganu and Penang, respectively. These chemicals included pesticides, cleaning detergents, perfumes and industrial materials (table 4). The GHS has classified them as hazardous chemicals; adverse health effects include dermatitis, hypersensitivity, inflammation, and eye and respiratory tract irritation. More hazardous chemicals were detected in Penang, which may explain the overall high prevalence of asthma symptoms compared with Johor Bahru.
This is the first study to use a high-throughput untargeted approach to profile indoor chemical exposure and asthma. Natural metabolites, including microbial and plant metabolites, were protective for asthma, and synthetic chemicals, including pesticides, detergents and industrial solvents, were risk factors for asthma. Also, this is the first study to assess the interactions between indoor microbiomes and metabolites, revealing that indoor microorganisms may produce protective metabolites. In addition, three centres in Malaysia with large geographic separation were surveyed to support our results and conclusions. We found large-scale indoor microbiome variation, indicating an overall different microbiome exposure in each centre. However, general rules were observed for indoor chemical compounds. Plant metabolites from flavonoids and isoflavonoids and microbial metabolites from indole derivatives showed potential protective effects, whereas synthetic chemicals showed adverse health effects. The results suggest that indoor chemicals could be a more solid and consistent indicator in exposure assessments for asthma. There are also limitations in this study. First, dust in Johor Bahru and Terengganu was collected from vacuuming dust from floors, desks, tables, bookshelves and curtains, whereas dust in Penang was collected from the blackboard frame, which may produce sampling bias. However, we argue that the two approaches should be comparable. Bookshelves, curtains and blackboard frames are seldom or never cleaned in these classrooms, and thus both approaches collect dust representing long-term exposure. Second, we applied second-generation amplicon sequencing in this study, only resolving taxonomic resolution at the genus level. Third, as only the marker gene is sequenced, the abundance and health associations for function genes cannot be assessed. However, our study profiled indoor microbial metabolites, which provide more direct evidence for metabolic exposures than functional gene assessment. Flavonoids and isoflavonoids are plant secondary metabolites with a polyphenolic structure. Flavonoids are found in many plants and plant-derived food, and isoflavonoids are predominantly found in soybeans and leguminous plants [29]. Flavonoids and isoflavonoids have anti-inflammatory, antioxidative, anticarcinogenic and antiallergic properties for humans and animals (table 3). Several flavonoids identified have shown protective effects for asthma in previous studies. For example, isoliquiritigenin suppresses interleukin (IL)-4 and IL-5 production in a dose-dependent manner in vitro [30], and astragalin reduces IL-4, IL-5 and IL-13 levels and inhibits eosinophil infiltration in mice [31]. Also, formononetin alleviates lung inflammation and cytokine levels and reduces oxidative stress in an ovalbumin-sensitised mouse model [32]. However, the previous studies mainly reported the health effects of flavonoids and isoflavonoids in laboratory animals and cell lines. This is the first study to show their beneficial effects as inhaled exposure for human populations, providing a novel perspective on respiratory disease. Indole and derivatives form a group of aromatic heterocyclic organic compounds widely distributed in bacteria, plants and animals [33]. The health effects of indoles were mainly studied in the human gut, with few studies on environmental microorganisms. A large variety of gut microorganisms can produce indole and derivatives, including Clostridium novyi, Escherichia coli, Fusobacterium, Enterococcus faecalis and Corynebacterium acnes [34, 35]. Indole and derivatives can improve human intestinal epithelial barrier integrity and reduce gut inflammation by decreasing the expression of pro-inflammatory cytokine NF-κB and increasing anti-inflammatory IL-10 [36]. Indolepyruvate and indole-3-acetamide were identified as potential protective metabolites in our study. In human gut studies, these indole derivatives can activate the expression of the aryl hydrocarbon receptor gene [37], which has an anti-inflammatory role in blocking pro-inflammatory T-cells in asthma development [38]. The inhaled exposure of indole metabolites could have a similar mechanism by activating aryl hydrocarbon receptors in the lung and respiratory tract. Besides flavonoids and indoles, biotin and chavicol were enriched in low-wheeze schools. Biotin, a B7 vitamin, is an essential nutrient for humans. A previous indoor metagenomics survey reported that a higher abundance of biotin metabolism pathways was associated with a lower prevalence of sick building syndrome [39]. Chavicol is a natural phenylpropene found in Piper betle, and is used in the traditional herbal medicine of China and India. Chavicol analogues can attenuate interferon-γ expression in T-helper cells, and modulate inflammation and immune responses [40], supporting their roles in reducing asthma symptoms. Only a few chemicals were significantly enriched in high-wheeze schools after removing drugs and common human and plant metabolites, and all of them were synthetic chemicals, including pesticides, paints, fragrances and industrial solvents. Phthalate exposure is reported to associate with asthma. A meta-analysis of 43 studies reported that benzyl phthalate increased the odds of childhood asthma by 39–41% [41]. A home survey in China also reported that a high concentration of phthalic acid esters increased childhood diagnosed asthma [42]. The other potential risk chemicals are not reported to associate with asthma by scientific publications, but the GHS classification, developed by the United Nations, indicated that these chemicals might have adverse health effects, including dermatitis, hypersensitivity and respiratory tract irritation. Thus, future environmental surveys and asthma epidemiology should also consider these chemicals. Previous indoor metabolite studies mainly surveyed microbial metabolites by low-throughput approaches. Araki et al. [12] and Choi et al. [43] reported that many MVOCs were positively associated with asthma and rhinitis. In our study, only three MVOCs (bornyl acetate, 2-heptanone and estragole) were detected in vacuum dust and none of them were significantly enriched in high/low-wheeze schools. It is likely that most volatile chemicals cannot be detected by vacuum dust sampling. The total LPS concentration seems to be mainly protectively associated with asthma [44]. Muramic acid was reported to be positively or negatively associated with asthma [45, 46]. In this study, the untargeted LC-MS also detected muramic acid and LPS (tridecanoic acid, hydroxy hexadecanoic acid and myristic acid), but none reached significance after the FDR adjustment.
In this study, we found large-scale variation in the microbiome composition and health-related microorganisms in three centres in Malaysia. This could be due to the extremely high diversity of environmental microorganisms [11]. Thus, it is challenging to use the indoor microbial composition to build a universal reference catalogue for health assessments and disease prediction. However, consistent associations were observed for indoor chemical compounds, suggesting they could be used as an environmental assessment indicator for disease prediction, providing new insights and strategies for disease prevention and control.
10.1183/13993003.00260-2022.Supp1 Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author. Supplementary tables ERJ-00260-2022.Tables Supplementary figures ERJ-00260-2022.Figures
10.1183/13993003.00260-2022.Shareable1 This one-page PDF can be shared freely online. Shareable PDF ERJ-00260-2022.Shareable |
|
PMC9647077 | Kalpana D. Acharya,Randall H. Friedline,Doyle V. Ward,Madeline E. Graham,Lauren Tauer,Doris Zheng,Xiaodi Hu,Willem M. de Vos,Beth A. McCormick,Jason K. Kim,Marc J. Tetel | Differential effects of Akkermansia-enriched fecal microbiota transplant on energy balance in female mice on high-fat diet | 27-10-2022 | gut microbiome,metabolism,estrogens,obesity,estradiol,diabetes | Estrogens protect against weight gain and metabolic disruption in women and female rodents. Aberrations in the gut microbiota composition are linked to obesity and metabolic disorders. Furthermore, estrogen-mediated protection against diet-induced metabolic disruption is associated with modifications in gut microbiota. In this study, we tested if estradiol (E2)-mediated protection against obesity and metabolic disorders in female mice is dependent on gut microbiota. Specifically, we tested if fecal microbiota transplantation (FMT) from E2-treated lean female mice, supplemented with or without Akkermansia muciniphila, prevented high fat diet (HFD)-induced body weight gain, fat mass gain, and hyperglycemia in female recipients. FMT from, and cohousing with, E2-treated lean donors was not sufficient to transfer the metabolic benefits to the E2-deficient female recipients. Moreover, FMT from lean donors supplemented with A. muciniphila exacerbated HFD-induced hyperglycemia in E2-deficient recipients, suggesting its detrimental effect on the metabolic health of E2-deficient female rodents fed a HFD. Given that A. muciniphila attenuates HFD-induced metabolic insults in males, the present findings suggest a sex difference in the impact of this microbe on metabolic health. | Differential effects of Akkermansia-enriched fecal microbiota transplant on energy balance in female mice on high-fat diet
Estrogens protect against weight gain and metabolic disruption in women and female rodents. Aberrations in the gut microbiota composition are linked to obesity and metabolic disorders. Furthermore, estrogen-mediated protection against diet-induced metabolic disruption is associated with modifications in gut microbiota. In this study, we tested if estradiol (E2)-mediated protection against obesity and metabolic disorders in female mice is dependent on gut microbiota. Specifically, we tested if fecal microbiota transplantation (FMT) from E2-treated lean female mice, supplemented with or without Akkermansia muciniphila, prevented high fat diet (HFD)-induced body weight gain, fat mass gain, and hyperglycemia in female recipients. FMT from, and cohousing with, E2-treated lean donors was not sufficient to transfer the metabolic benefits to the E2-deficient female recipients. Moreover, FMT from lean donors supplemented with A. muciniphila exacerbated HFD-induced hyperglycemia in E2-deficient recipients, suggesting its detrimental effect on the metabolic health of E2-deficient female rodents fed a HFD. Given that A. muciniphila attenuates HFD-induced metabolic insults in males, the present findings suggest a sex difference in the impact of this microbe on metabolic health.
Loss of estrogens during menopause causes weight gain, resulting in an increased risk of metabolic, cardiac, inflammatory, osteopathic, and neurological disorders (1–4). Protection against diet-induced obesity, hyperglycemia, hyperlipidemia, and insulin resistance are mediated by estrogens in women and in female rodents (5–9). E2-dependent protection against HFD-induced obesity is associated with increased physical activity and basal energy expenditure and improvements in systemic insulin sensitivity and glucose metabolism (7, 9). Gut microbiota profoundly impact host metabolism. Metabolic syndrome, characterized by adiposity, hyperlipidemia and hyperglycemia, is associated with changes in gut microbiota (10–12). Manipulation of gut microbiota, by depletion via antibiotics (12–16), administration of specific bacteria (17–20), or transplantation of fecal/caecal microbiota (14, 21, 22) can improve metabolism, in male rodents and men. In particular, the abundance of bacteria belonging to the Verrucomicrobia and its dominant intestinal genus Akkermansia, are negatively associated with obesity in men and women (23–26). Akkermansia muciniphila is a mucin utilizer and producer of short chain fatty acids, which have anti-inflammatory properties and are primary nutrients for intestinal endocrine cells (27–29). Moreover, A. muciniphila produces signaling proteins, such as the 33-kD Amuc_1100, that interact with the Toll-like 2 receptor and improve barrier function (19, 30). In male mice fed a HFD, A. muciniphila supplementation attenuated obesity and inflammation and improved insulin signaling (17, 19, 31, 32). Only a few studies have investigated the effects of Akkermansia in women. In postmenopausal women, Akkermansia was negatively correlated with insulin resistance and dyslipidemia (33). Furthermore, heat-killed A. muciniphila administration attenuated body weight, fat mass, and hip circumference in obese women and men (18). There is increasing evidence that estrogens can influence gut microbiota (34–36). Postmenopausal women have higher Prevotella and lower Lachnospira and Roseburia relative abundances, and a lower Firmicutes/Bacteroidetes ratio, when compared to premenopausal women (37). The differences in gut microbiota were attenuated between postmenopausal women and men, and between gonadectomized male and female rats, although baseline sex differences in gut microbiota persist even after the depletion of gonadal estrogens (37, 38). It is possible that these sex differences start during puberty as girls were found to develop towards an adult microbiota earlier than boys (39). Intake of phytoestrogens in women was found to increase beneficial microbes including Lactobacillus, Enterococcus and Bifidobacterium (40, 41). Ovariectomy or E2 treatment, in wild-type as well as ob/ob (leptin-deficient) mice, altered gut microbiota (7, 42–45). Interestingly, while HFD or a high-fat high-sugar diet decreased relative Akkermanisa levels in male mice (17, 31, 46), these were increased in HFD-treated female mice, with a further increase in E2-treated groups (7). This sex difference in Akkermansia modulation in response to a change in diet indicates a critical need to examine the functions of gut microbiota, including Akkermansia supplementation, on female metabolic health. Therefore, in this study, using female mice, we tested if fecal microbiota transplantation (FMT) from E2-treated lean mice, with or without A. muciniphila supplementation, protects E2-deficient mice against HFD-induced metabolic insults.
Animal experiments were performed at the University of Massachusetts Chan Medical School and Wellesley College. All procedures were approved by the Institutional Animal Care and Use Committees of University of Massachusetts Chan School and Wellesley College and performed in accordance with National Institutes of Health Animal Care and Use Guidelines.
Ten-week-old female C57BL/6J mice were housed 3-4/cage on a 12h light-dark cycle, with ad libitum food and water. Mice were ovariectomized and silastic capsules filled with 17β-estradiol (E2, 50 μg in 25 μl of 5% ethanol/sesame oil), or vehicle (Veh, 25 μl of 5% ethanol/sesame oil) were subcutaneously implanted as described previously (6, 7, 44, 47).
Both donor and recipient female mice were fed phytoestrogen-free standard chow (13% kCal from fat, LabDiet, #5V75R) until they were switched to HFD containing 60% kCal fat (#D12492, Open Sources Diet, USA) for the remainder of the study.
To allow an efficient colonization of the donor microbiota by initially depleting native microbiota, all recipients were administered an antibiotic cocktail of ampicillin (1 g/L; #A0166, Sigma-Aldrich, USA), vancomycin (500 mg/L; #PHR1732, Sigma-Aldrich, USA) neomycin (1 g/L; #N6386, Sigma-Aldrich, USA), and metronidazole (1 g/L; #M3761, Sigma-Aldrich, USA) (AVNM), as described previously (48), for a total of 9 days in drinking water. This antibiotic cocktail regimen has been shown to be effective in reducing up to 90% of the native bacterial community and depletes most groups of microbes (e.g., gram positive, gram negative and anaerobes) in male and female mice (49–51).
Fresh fecal samples from mice of the same treatment group were collected and pooled on the morning of the gavage, as described previously (48). FMT was diluted in PBS buffer (0.01M) reduced with 0.5% L-cysteine HCl (1:10), in an anaerobic chamber. Anaerobic environment was created by purging with gas mix (5% H2/10% CO2/85% N) 2-3 times, until the chamber gas reading was 3% H2 and 0 ppm O2. Up to 500µL of reduced PBS buffer was added to 10-15 fecal pellets and gently lysed until no visible pieces were present. FMT mixture was filtered using 70 µm mesh filter and diluted in PBS buffer to bring to a final concentration of 100 mg/ml. Following antibiotic treatment, recipients orally gavaged with 150 µL of 100mg/mL FMT (52, 53). FMT was started on D10 and continued for a total of 9 doses ( Figure 1A ). The days of FMT for the recipients were matched to the days of the fecal sample collection from the donors (e.g. on D12, the recipients received FMTs from fecal samples that were collected on D12 from the donors). V-V and V-E recipients received FMT gavage from Veh and E2 donors, respectively.
Body weight and body composition (lean/fat mass, using 1H-MRS) were measured throughout the study ( Figure 1A ). Blood glucose levels during a five-hour fasting period were measured during chow feeding (D9), a week after the start of HFD (D19) and at the end of the study (D43), to assess the effect of transplanted gut microbiota on glucose homeostasis.
The effects of E2 and FMT on HFD-induced longitudinal metabolic changes, starting on D10 (1st FMT gavage day), including body weight, fat mass, and lean mass were separately analyzed by a two-way repeated measures ANOVA, followed by a Student’s t-test for the days when an effect was present (Jamovi, v 1.8.4.0). A two-way ANOVA followed by a Tukey’s HSD post-hoc was used to measure the effects of E2 and FMT on blood glucose levels across groups.
Recipient mice were treated as described above except that the AVNM cocktail was administered for 14 days (49, 51, 54), and HFD was introduced on D14 ( Figure 2A ).
A muciniphila MucT (ATTC BAA-835) cells were grown in a synthetic medium containing 16 g/l soy-peptone, 4 g/l threonine, and a mix of glucose and N-acetylglucosamine (25 mM each) under strictly anaerobic conditions (19). Cells were then washed in reduced PBS with 25% (vol/vol) glycerol and immediately frozen at −80°C. Within two hours prior to the gavage, fecal pellets were lysed in reduced 0.01M PBS buffer (containing 0.05% L-cysteine HCl) in anaerobic chamber, as described above in Experiment 1 (19). For transfer to the experimental laboratory, the cells were shipped in dry ice, and upon receipt, were quickly aliquoted in smaller volumes for daily gavages on ice under strict anaerobic conditions and stored at -80 C. Within two hours prior to each gavage, FMT were prepared as described above. A. muciniphila preparations were thawed on ice and immediately mixed with the fresh FMT. V-EA recipients (n=4) received 150 μl of oral gavage containing 40mg/mL of FMT supplemented with 2 × 108 A. muciniphila. A. muciniphila-supplemented FMT were started on D15 and gavaged every other day for a total of 6 doses ( Figure 2A ). Control V-V (n=4) and E-E (n=4) mice were similarly gavaged with FMT from Veh or E2 mice, respectively, without A. muciniphila cells supplementation.
To assess the HFD-induced metabolic changes as an effect of FMT from lean E2-treated mice supplemented with A. muciniphila, food and water intake, respiration, energy expenditure, and locomotor activity were measured in awake mice after 10 days on HFD (D28-D31; Figure 2A ) using metabolic cages (TSE Systems, Germany), as described previously (7, 55). Resting energy expenditure (EE) and respiratory exchange ratio (RER) were derived from O2 consumption and CO2 production data. Body weight and body composition (lean/fat mass) were measured throughout the study using 1H-MR spectroscope (EchoMRI, Houston, TX, USA).
Fasting blood glucose was measured weekly starting with HFD feeding and FMT administration. On day 45 following overnight fasting, a glucose tolerance test (GTT) was performed to measure insulin sensitivity. In brief, 20% glucose at 1g/kg BW was injected i.p. and glucose measurements were taken at 0, 15, 30, 60, 90, and 120 mins following injection. Mice were euthanized immediately following GTT.
Fecal samples were collected from donor and recipient mice throughout the study to confirm the microbial transfer in recipients and to examine the association between gut microbiota and host metabolism. Fresh fecal samples were collected, immediately frozen in dry ice, and stored at -80°C. Total DNA was extracted from fecal pellets using the DNeasy PowerSoil Kit (Cat #12888, Qiagen, USA) following the manufacturer’s protocol.
Microbiome community profiling of fecal DNA was performed by 300nt paired-end 16S rRNA gene sequencing of the V3-4 region on the Illumina MiSeq platform as described (56). The UPARSE/SINTAX pipeline (usearch v10.0.240_i86linux6, rdp_16s_v18.fa) (57) was used to define OTUs and assign taxonomic classifications.
To examine the combined effects of Akkermansia enrichment and FMT from lean E2-treated mice on HFD-induced metabolic changes, longitudinal data starting on D17 (after the first FMT and the start of HFD), including body weight, fat mass, lean mass, blood glucose, and glucose tolerance test (GTT) were analyzed, as described above in Experiment 1. Data from metabolic cage experiments were analyzed using one way ANOVA, or ANCOVA using body weight immediately prior to the metabolic cages as a covariate (VO2 and VCO2), followed by a Tukey posthoc test.
Prior to analysis, OTUs which failed to classify to at least the taxonomic Family level were removed to reduce spurious OTUs. Abundances were summed according to assigned taxonomic classifications for analysis at higher taxonomic levels. Analysis was conducted using either QIIME2 (ver. 2021.4) (58) or MaAsLin2 (ver. 1.8.0) (59) as appropriate. For multiple comparisons, FDR corrections were done and q<0.1 was considered significant.
In order to assess the effects of E2 on body composition, ovariectomized mice receiving implants of estradiol (E2, n=7) or vehicle (Veh, n=7), were analyzed for changes in body weight and fat mass. In contrast to the recipient mice (below), these animals served as E2 donors and Veh donors, and did not receive antibiotic treatment. E2 prevented weight gain over 6 weeks ( Figure 1B ). Veh groups weighed more than E2-treated mice after 3 days on HFD (D14) through the end of the study ( Figure 1B ). Moreover, E2 attenuated fat mass gain compared to Veh mice (E-donor and V-donor, respectively; Figure 1C ). As reported previously (7) lean mass was not affected by E2 treatment (data not shown). E2 also prevented hyperglycemia in mice fed HFD for a week ( Figure 1D ), with this effect maintained after 4 weeks on HFD (after 6 weeks of E2 implant; T-test, p<0.001; 95%CI [-77.3, -40.4].
Ten-week old ovariectomized C57BL/6J recipient mice were divided into 2 groups: 1) Veh implanted recipients that received FMT from V-donor mice (V-V, n=5) and 2) Veh implanted recipients that received FMT from E2-treated donor mice (V-E, n=6). In addition to the FMT, V-V and V-E recipients were cohoused with Veh and E2 mice, respectively, at a ratio of 1:1 to transfer microbiota via coprophagy (60), starting on the last day of antibiotic treatment ( Figure 1A ). FMT did not protect recipient against HFD-induced body weight and fat mass gain ( Figures 1B, C ). While FMT and cohousing with E2-treated lean mice did not affect blood glucose levels, a trend (p=0.07, one-tailed t-test) towards a decrease was detected in V-E recipients compared to V-V on D19 ( Figure 1D ).
Ten week-old female ovariectomized C57BL/6J E2- or Veh-implanted mice (n=7/group)were group-housed with 3 mice/cage. As in the Experiment 1, the FMT gavage days in recipients were matched with the fecal sample collection days in donors. Recipients were divided into 3 groups with: 1) Veh implants receiving FMT from Veh mice (V-V; n=4), 2) Veh implants receiving FMT from E2-treated mice supplemented with A. muciniphila cells immediately prior to gavage (V-EA; n=4) and 3) E2 implants receiving FMT from E2-treated mice (E-E; n=4) ( Figure 2A ). FMT from E2-treated lean mice supplemented with A. muciniphila did not prevent body weight and fat mass gain in ovariectomized mice. E-E mice gained less body weight ( Figure 2B ) and fat mass ( Figure 2C ) compared to both Veh groups, V-V and V-EA. Longitudinal analysis showed that E-E mice weighed less than V-EA mice from D23 and through the rest of the study ( Figure 2B ). Similarly, E-E mice weighed much less than V-V mice, on D27, D29 and D32-D35 (p<0.05; Figure 2B ). Compared to the E-E group, the weight gain in the V-V and V-EA mice was mostly due to fat weight starting on D25 ( Figure 2C ). Unlike E2 treatment, FMT from E2-treated lean mice supplemented with A. muciniphila did not prevent weight gain or fat mass in recipient females. While the body weight of V-EA mice tended to be slightly higher than V-V controls a week after the last FMT ( Figure 2 ), this effect was not significant (p=0.16). As in Experiment 1 described above, lean mass was not affected by E2 treatment or FMT.
Fasting blood glucose levels were measured weekly in recipient mice during HFD feeding. E-E mice had lower blood glucose compared to both V-EA and V-V groups on D25, and V-EA on D32 ( Figure 3A ), indicating protection from HFD-induced hyperglycemia. To assess the effects of E2 on insulin sensitivity, GTT was measured in recipients following overnight fasting and injection of 20% glucose (i.p., 1g/kg body weight) (61). GTT blood glucose was increased in both V-V and V-EA groups compared to E-E mice at 90 and 120 mins following injection ( Figure 3B ). Unlike E2 treatment, FMT from E2 mice supplemented with A. muciniphila increased fasting glucose levels in V-EA compared to V-V controls on D25 (p<0.05, t-test; Figure 3A ), suggesting a negative impact of FMT supplemented with Akkermansia on glucose homeostasis in female mice. However, A. muciniphila-supplemented FMT had no effect on GTT glucose levels ( Figure 3B ).
To test if E2-mediated protection against HFD-induced obesity is associated with energy intake, food and water consumption were measured in female mice, using metabolic cages. E-E groups ate less during 24h and showed a strong trend towards a decrease during night (p=0.062), compared to V-EA groups. E2 increased physical activity during night in E-E mice compared to V-V and V-EA mice. In all treatment groups, food intake and physical activity peaked at 2h after light-off and an hour before the light-on phase ( Supplemental Figure 1 ). Additionally, E2 increased basal energy expenditure and VO2 consumption and showed a strong trend towards an increase in VCO2 production (p=0.06) during night, compared to V-V groups ( Figure 4 ). Similarly, V-EA groups showed a slight trend towards an increase in energy expenditure compared to E-E mice (p=0.08). In contrast to E2, FMT from E2-treated animals supplemented with A. muciniphila did not improve any metabolic measures. Taken together, these data suggest that E2 regulates energy homeostasis, in part by decreasing energy intake and increasing energy expenditure, whereas FMT has no effect.
Fecal DNA was used to generate 16S rRNA amplicons that were sequenced at 0, 1 and 2 weeks to examine the effect of the two week-long antibiotic treatment. A longitudinal analysis of gut microbiota during antibiotic treatment showed an effect of time (F(2,20)=15, p=0.036) on diversity (Faith’s PD (62). Specifically, D14 diversity differed from D1 (ANOVA, p=0.038) and showed a trend towards a decrease compared to D7 (ANOVA, p=0.06; Figure 5A ). Similarly, β-diversity (weighted UniFrac), showed that the D1 microbiome was significantly different from D7 and D14 (q=0.001; pairwise PERMANOVA) ( Figure 5B ; Table 1 ), where the effect of antibiotic was explained by principal component 1(67%). Taken together, these results confirm that antibiotics deplete microbial community as early as within one week of the treatment. The effect of antibiotics on taxonomic abundances was examined by combining the three treatment groups within each day for D1-D14 data and comparing across days. Multiple Clostridium species, Oscillibacter, Coprococcus, Anaerotruncus and Eubacterium were decreased on both D7 and D14 compared to D1. An additional 12 taxa, including Bacteroides, Anaeroplasma, Turicibacter, and Acetatifactor, were decreased on D14, compared to D1 as an effect of antibiotic treatment ( Table 1 ; Figure 6 ). Aberrations in the fecal communities, including increased levels of multiple Bacillus, Rhizobium, Pantoea, Corynebacterium, and Lactococcus spp., were observed due to antibiotic treatment on D7 and D14. Thirteen additional taxa, including Akkermansia, Bifidobacterium, Romboutsia and Devosia, were increased on D7, compared to D1. Other taxa, such as Prevotella, Fecalibacterium, Mycobacterium, and Methalobacterium, were increased on D14 only compared to D1 ( Table 1 ; Figure 6 ).
To examine the effects of FMT during chow or HFD feeding, gut microbiota was compared between treatment groups within each diet. First, the presence of Akkermansia was confirmed in the FMT samples that were supplemented with A. muciniphila before the gavage ( Supplemental Figure 2 ). To determine the effects of FMT during chow, gut microbiota from D17 was analyzed. Gut microbiota was profoundly altered on D17, two days after a single dose of the A. muciniphila -supplemented FMT in the V-EA group. As expected, this change in microbiota at D17 was primarily due to A. muciniphila, with about 30% of the total gut microbiota in V-EA recipients being comprised of A. muciniphila ( Figure 6B ). While the α-diversity (Faith PD) did not differ across groups, microbial composition clustered separately between V-EA and V-V mice (PERMANOVA, q=0.037; Figure 6A and Table 2 ). The profound increase in A. muciniphila in V-EA mice on D17 was accompanied with decreases in Clostridium_sensu, Parasutterella, and Bacteroides, suggesting an increase in A. muciniphila in antibiotic-treated mice can negatively impact abundance of other microbes in the gut ( Table 2 ). Once mice were started on HFD, relative abundance data from the FMT on days 23 and 27 were aggregated to represent the effect during FMT. V-EA recipients had an increase in phylogenetic diversity compared to E-E (ANOVA, Tukey posthoc, p=0.009), but were similar to V-V. Additionally, D23 and D27 microbial community of V-EA mice clustered separately from E-E groups (PERMANOVA, q=0.004; Figure 6A ). The effect of the FMT was detected on taxa levels as well, such that V-EA mice had higher relative abundances of Anaerotruncus and Erysipelotrichaceae_incertae_sedis compared to V-V controls ( Figure 6B ; Table 2 ). Of interest, the Akkermansia levels increased as expected in the D17 sample of the V-EA mice, consistent with our earlier finding in females (7) but different from findings reported in males (46). Gut microbiota from the two weeks (D32 and D39) immediately after the last FMT treatment were aggregated and analyzed to capture the late-emerging effects of FMT. In the two weeks post-FMT, V-EA mice had increased α-diversity compared to V-V (p=0.019) and E-E (p<0.001) ( Figure 5A ). Moreover, V-EA microbial community distances differed from both V-V and E-E groups, (PERMANOVA, q=0.003) ( Figure 6A ), suggesting a long-lasting effect of FMT treatment on gut microbiota. On taxa level, V-EA mice showed increased abundances of Oscillibacter and Desulfovibrio compared to V-V ( Table 2 and Figure 6B ), suggesting that the effect of FMT continued even after the treatment was ended.
On D17 (a week after E2 treatment), E-E groups clustered differently from V-V and V-EA (PERMANOVA, q=0.037, Figure 6A ). These differences in communities were due to decreases in relative abundances of Turicibacter, Parasutterella, Enterococcus, and Clostridium_sensu_stricto, in E-E mice compared to V-V ( Figure 6B and Table 3 ). The effect of E2 on driving differential clustering continued after switching to HFD, as shown by aggregate data on D23 and D27. E-E groups clustered separately from V-V (PERMANOVA, q=0.003, Figure 6A ), although α-diversity was not affected by E2. These changes were mostly due to increases in relative abundances of Escherichia. Shigella and Akkermansia, and decreases in that of Anaerostipes, Turicibacter, Lactococcus, Lactobacillus, Blautia and Clostridium_ IV/XIVa/XIVb ( Table 3 and Figure 6B ). Similarly, E2 altered both α-diversity and β-diversity on D32 and D39 (aggregate data) in HFD-fed mice. E-E groups had decreased α-diversity compared to V-V mice (ANOVA, p=0.035, Figure 5A ). On the community level, E-E mice clustered differently from V-EA (PERMANOVA, q=0.042), but not V-V ( Figure 6A ), suggesting that the effect of E2 on gut microbiota started to attenuate around 4th week of E2 implant. During this time, E-E mice had increased relative Escherichia.Shigella and Parasutterella abundances compared to V-V groups. In contrast, E-E mice had lower relative abundances of Acetanaerobacterium, Anaerotruncus, Clostridium_XVIII/XIV, Turicibacter, Enterococcus, Lactobacillus, and Romboutsia, compared to V-V mice, on D32 and D39 ( Table 3 and Figure 6B ). Interestingly, the E2-induced increase in the relative abundance of Akkermansia observed on the D23 and D27 of E2 treatment did not persist at D32 or D39, suggesting that HFD increased Akkermansia in both groups, eliminating the difference between the E2 and Veh groups.
In the present study, we tested the hypothesis that the gut microbiota mediates some of the protective effects of estrogens on energy metabolism in female mice. Using adult female mice, we investigated the metabolic outcome of cohousing and transfer of the gut microbiota from estrogen-treated lean donors to estradiol-deficient HFD-fed mice. The present findings extend previous reports that estradiol treatment protects ovariectomized HFD-fed mice from hyperphagia, obesity, and hyperglycemia and improves active and basal energy expenditure (5–9, 44). Contrary to our hypothesis, we found that FMT from E2-treated lean donors was not sufficient to transfer a lean phenotype and metabolic benefits to ovariectomized recipients fed a HFD, although a tendency towards improved blood glucose levels was present. In an effort to maximize the potential effects of gut microbiota on metabolism in female mice, we supplemented the FMT from lean E2-treated mice with A. muciniphila, a bacterial species previously reported to alleviate metabolic insults in male rodents, men, and women (17–19, 29, 31–33, 63), but c.f ( 64 ). In addition, we have previously identified that the relative abundance of Akkermansia increases in E2-treated female mice compared to ovariectomized controls and is inversely correlated to weight gain and fat mass (44). In the current study, while an FMT supplemented with A. muciniphila altered the relative abundance of many gut microbial species, surprisingly, it did not improve metabolic health but seemed to negatively affect blood glucose in ovariectomized mice fed HFD. These novel findings in female mice suggest that a transplant of fecal microbiota supplemented with A. muciniphila, under the present experimental conditions, is not sufficient to transfer the metabolic phenotype and could aggravate some HFD-induced insults. Although the colonization of many microbes via FMT persisted, introduction of HFD profoundly and acutely increased Akkermansia in all treatment groups. These data suggest that enriching FMT from lean mice with A. muciniphila disrupts glucose homeostasis. Alternatively, while the supplementation with only A. muciniphila could have beneficial effects on metabolic health, addition of A. muciniphila in feces from E2-treated female mice, that already contain this microbe, may exert detrimental effects by disrupting the microbial community homeostasis. The different effects of Akkermansia between the current study and previous findings may be due to sex differences in the effects of Akkermansia in the mammalian gut. Neither HFD nor a high-fat high-sugar diet induced Akkermansia in male mice (17, 31, 46). In dramatic contrast, a week of HFD in female mice elicited a robust increase in the relative abundance of Akkermansia as reported here and previously (7). Akkermansia uses mucin as its sole nutrient source, which is an integral component of the gastrointestinal mucosa layer (25, 28). Sex differences also exist in intestinal mucin in the manifestation of obesity. In male mucin2 knockout mice, HFD-induced obesity and hyperglycemia from alcohol-induced hepatosteatosis was attenuated (65, 66), while female mucin2-deficient mice had exacerbated glucose tolerance and were not protected from obesity. Thus, a sex difference in the nutrient source for Akkermansia could lead to differences in their abundance and function between males and females (Hartmann, 2016). In addition, but not mutually exclusive of this possible sex difference, the disparate outcomes between the present findings and previous studies could be due to other differences in experimental design, including use of cecal vs. fecal material transplant, use of live vs. killed A. muciniphila cells, the number and frequency of FMT gavages, and the presence or absence of cohousing donors (14, 17, 19). It is important to note that gut microbiota elicits a variety of responses based on the factors contributing to metabolic disorders. For example, a transfer of healthy microbiota attenuated body weight gain and improved insulin response in PCOS models of female mice) (60), whereas did not prevent ovariectomy-dependent obesity (67). Most importantly, the present findings provide a compelling justification for further investigation of sex differences in basic and clinical studies in the function of gut microbiota in metabolic health. The differences between the present and previous studies could also be due to the use of antibiotics for initial depletion of the native gut microbiota prior to A. muciniphila gavage in the current study unlike in previous studies (17, 19). Antibiotics interact with estrogens (34, 68, 69), primarily by altering the composition of the gut microbiota and E2 metabolism. In support, the mammalian gut is ubiquitously colonized by microbes that produce the steroid-metabolizing enzyme, β-glucuronidase, which is responsible for the deconjugation and reuptake of E2 in enterohepatic circulation (34, 70–72). β-glucuronidase activity has been observed in Bacteroides and Ruminococcus (73), which were decreased by antibiotics in the present study. Antibiotic treatment decreases β-glucuronidase and increases excretion of conjugated estrogens in feces (71, 74). Thus, the reuptake and availability of E2 was likely diminished by antibiotic administration in the first two weeks of the study, possibly via depletion of this E2-metabolizing microbial community. Although the goal of using the lean-FMT background for A. muciniphila in the current study was to replenish the beneficial microbial community depleted by antibiotics, it is possible that the HFD intake after the antibiotic treatment permanently disrupted the healthy microbial ecosystem, potentially resulting in an increased mucus production that may have resulted in an overwhelming increase of Akkermansia. In this context, it is of interest to note that blooms of Akkermansia spp., in human have been described in antibiotic-treated male patients without apparent negative health effects (75). In the current study, FMT from E2-treated lean mice supplemented with A. muciniphila caused hyperglycemia and an increased trend towards body weight gain, suggesting A. muciniphila supplementation has a detrimental effect on metabolic health in female mice under the used conditions. The increase in Akkermansia following gavage with A. muciniphila-supplemented FMT was accompanied by a parallel decrease in Parasutterella and Bacteroides. Parasutterella is decreased in prediabetic rats and pregnant women with gestational diabetes mellitus and this decrease is associated with a decrease in short chain fatty acid levels (76–78). Similarly, administration of Bacteroides attenuates HFD-induced obesity, hyperglycemia and insulin resistance in rats (79). Taken together with the present study, these findings suggest that a decrease in these microbes contributes to Akkermansia-dependent impairment in metabolism in females. In summary, FMT from lean E2-treated mice mildly improves blood glucose levels in female mice fed a HFD, but does not protect from obesity. However, enriching FMT from lean mice with A. muciniphila disrupts glucose homeostasis in the present model. Based on the existing evidence of beneficial effects of A. muciniphila on metabolic health, mostly observed in male animal models, clinical trials using an A. muciniphila supplement in humans have been completed (18, 19), where safety and efficacy of A. muciniphila supplementation have been shown in both female and male subjects. It will be important to determine if the present findings in estradiol-deficient HFD female mice extrapolate to humans. Moreover, it is critical that future studies investigate sex differences in host-Akkermansia interactions regarding metabolic health.
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: <b><br>https://www.ncbi.nlm.nih.gov/, SAMN30108647 C01 C01 mouse metagenome 1441287 SAMN30108648 C12 C12 mouse metagenome 1441287 SAMN30108649 C13 C13 mouse metagenome 1441287 SAMN30108650 C02 C02 mouse metagenome 1441287 SAMN30108651 C03 C03 mouse metagenome 1441287 SAMN30108652 C04 C04 mouse metagenome 1441287 SAMN30108653 C07 C07 mouse metagenome 1441287 SAMN30108654 C08 C08 mouse metagenome 1441287 SAMN30108655 EE19 EE19 mouse metagenome 1441287 SAMN30108656 EE21 EE21 mouse metagenome 1441287 SAMN30108657 EE23 EE23 mouse metagenome 1441287 SAMN30108658 EE25 EE25 mouse metagenome 1441287 SAMN30108659 EE27 EE27 mouse metagenome 1441287 SAMN30108660 D01EEM10 D01EEM10 mouse metagenome 1441287 SAMN30108661 D14EEM10 D14EEM10 mouse metagenome 1441287 SAMN30108662 D17EEM10 D17EEM10 mouse metagenome 1441287 SAMN30108663 D23EEM10 D23EEM10 mouse metagenome 1441287 SAMN30108664 D27EEM10 D27EEM10 mouse metagenome 1441287 SAMN30108665 D32EEM10 D32EEM10 mouse metagenome 1441287 SAMN30108666 D39EEM10 D39EEM10 mouse metagenome 1441287 SAMN30108667 D07EEM10 D07EEM10 mouse metagenome 1441287 SAMN30108668 D01EEM11 D01EEM11 mouse metagenome 1441287 SAMN30108669 D14EEM11 D14EEM11 mouse metagenome 1441287 SAMN30108670 D17EEM11 D17EEM11 mouse metagenome 1441287 SAMN30108671 D23EEM11 D23EEM11 mouse metagenome 1441287 SAMN30108672 D27EEM11 D27EEM11 mouse metagenome 1441287 SAMN30108673 D32EEM11 D32EEM11 mouse metagenome 1441287 SAMN30108674 D39EEM11 D39EEM11 mouse metagenome 1441287 SAMN30108675 D07EEM11 D07EEM11 mouse metagenome 1441287 SAMN30108676 D01EEM12 D01EEM12 mouse metagenome 1441287 SAMN30108677 D14EEM12 D14EEM12 mouse metagenome 1441287 SAMN30108678 D17EEM12 D17EEM12 mouse metagenome 1441287 SAMN30108679 D23EEM12 D23EEM12 mouse metagenome 1441287 SAMN30108680 D27EEM12 D27EEM12 mouse metagenome 1441287 SAMN30108681 D32EEM12 D32EEM12 mouse metagenome 1441287 SAMN30108682 D39EEM12 D39EEM12 mouse metagenome 1441287 SAMN30108683 D07EEM12 D07EEM12 mouse metagenome 1441287 SAMN30108684 D01VVM01 D01VVM01 mouse metagenome 1441287 SAMN30108685 D14VVM01 D14VVM01 mouse metagenome 1441287 SAMN30108686 D17VVM01 D17VVM01 mouse metagenome 1441287 SAMN30108687 D23VVM01 D23VVM01 mouse metagenome 1441287 SAMN30108688 D27VVM01 D27VVM01 mouse metagenome 1441287 SAMN30108689 D32VVM01 D32VVM01 mouse metagenome 1441287 SAMN30108690 D39VVM01 D39VVM01 mouse metagenome 1441287 SAMN30108691 D07VVM01 D07VVM01 mouse metagenome 1441287 SAMN30108692 D14VVM02 D14VVM02 mouse metagenome 1441287 SAMN30108693 D17VVM02 D17VVM02 mouse metagenome 1441287 SAMN30108694 D23VVM02 D23VVM02 mouse metagenome 1441287 SAMN30108695 D27VVM02 D27VVM02 mouse metagenome 1441287 SAMN30108696 D32VVM02 D32VVM02 mouse metagenome 1441287 SAMN30108697 D39VVM02 D39VVM02 mouse metagenome 1441287 SAMN30108698 D07VVM02 D07VVM02 mouse metagenome 1441287 SAMN30108699 D01VVM03 D01VVM03 mouse metagenome 1441287 SAMN30108700 D14VVM03 D14VVM03 mouse metagenome 1441287 SAMN30108701 D17VVM03 D17VVM03 mouse metagenome 1441287 SAMN30108702 D23VVM03 D23VVM03 mouse metagenome 1441287 SAMN30108703 D27VVM03 D27VVM03 mouse metagenome 1441287 SAMN30108704 D32VVM03 D32VVM03 mouse metagenome 1441287 SAMN30108705 D39VVM03 D39VVM03 mouse metagenome 1441287 SAMN30108706 D07VVM03 D07VVM03 mouse metagenome 1441287 SAMN30108707 D01VVM04 D01VVM04 mouse metagenome 1441287 SAMN30108708 D14VVM04 D14VVM04 mouse metagenome 1441287 SAMN30108709 D17VVM04 D17VVM04 mouse metagenome 1441287 SAMN30108710 D23VVM04 D23VVM04 mouse metagenome 1441287 SAMN30108711 D27VVM04 D27VVM04 mouse metagenome 1441287 SAMN30108712 D32VVM04 D32VVM04 mouse metagenome 1441287 SAMN30108713 D39VVM04 D39VVM04 mouse metagenome 1441287 SAMN30108714 D07VVM04 D07VVM04 mouse metagenome 1441287 SAMN30108715 D01VAM05 D01VAM05 mouse metagenome 1441287 SAMN30108716 D14VAM05 D14VAM05 mouse metagenome 1441287 SAMN30108717 D17VAM05 D17VAM05 mouse metagenome 1441287 SAMN30108718 D23VAM05 D23VAM05 mouse metagenome 1441287 SAMN30108719 D27VAM05 D27VAM05 mouse metagenome 1441287 SAMN30108720 D32VAM05 D32VAM05 mouse metagenome 1441287 SAMN30108721 D39VAM05 D39VAM05 mouse metagenome 1441287 SAMN30108722 D07VAM05 D07VAM05 mouse metagenome 1441287 SAMN30108723 D01VAM06 D01VAM06 mouse metagenome 1441287 SAMN30108724 D14VAM06 D14VAM06 mouse metagenome 1441287 SAMN30108725 D17VAM06 D17VAM06 mouse metagenome 1441287 SAMN30108726 D27VAM06 D27VAM06 mouse metagenome 1441287 SAMN30108727 D32VAM06 D32VAM06 mouse metagenome 1441287 SAMN30108728 D39VAM06 D39VAM06 mouse metagenome 1441287 SAMN30108729 D07VAM06 D07VAM06 mouse metagenome 1441287 SAMN30108730 D01VAM07 D01VAM07 mouse metagenome 1441287 SAMN30108731 D14VAM07 D14VAM07 mouse metagenome 1441287 SAMN30108732 D17VAM07 D17VAM07 mouse metagenome 1441287 SAMN30108733 D23VAM07 D23VAM07 mouse metagenome 1441287 SAMN30108734 D27VAM07 D27VAM07 mouse metagenome 1441287 SAMN30108735 D32VAM07 D32VAM07 mouse metagenome 1441287 SAMN30108736 D39VAM07 D39VAM07 mouse metagenome 1441287 SAMN30108737 D07VAM07 D07VAM07 mouse metagenome 1441287 SAMN30108738 D01VAM08 D01VAM08 mouse metagenome 1441287 SAMN30108739 D14VAM08 D14VAM08 mouse metagenome 1441287 SAMN30108740 D17VAM08 D17VAM08 mouse metagenome 1441287 SAMN30108741 D23VAM08 D23VAM08 mouse metagenome 1441287 SAMN30108742 D27VAM08 D27VAM08 mouse metagenome 1441287 SAMN30108743 D32VAM08 D32VAM08 mouse metagenome 1441287 SAMN30108744 D39VAM08 D39VAM08 mouse metagenome 1441287 SAMN30108745 D07VAM08 D07VAM08 mouse metagenome 1441287 SAMN30108746 D01EEM09 D01EEM09 mouse metagenome 1441287 SAMN30108747 D14EEM09 D14EEM09 mouse metagenome 1441287 SAMN30108748 D17EEM09 D17EEM09 mouse metagenome 1441287 SAMN30108749 D23EEM09 D23EEM09 mouse metagenome 1441287 SAMN30108750 D27EEM09 D27EEM09 mouse metagenome 1441287 SAMN30108751 D32EEM09 D32EEM09 mouse metagenome 1441287 SAMN30108752 D39EEM09 D39EEM09 mouse metagenome 1441287 SAMN30108753 D07EEM09 D07EEM09 mouse metagenome 1441287 SAMN30108754 VA19 VA19 mouse metagenome 1441287 SAMN30108755 VA21 VA21 mouse metagenome 1441287 SAMN30108756 VA23 VA23 mouse metagenome 1441287 SAMN30108757 VA25 VA25 mouse metagenome 1441287 SAMN30108758 VA27 VA27 mouse metagenome 1441287 SAMN30108759 VV19 VV19 mouse metagenome 1441287 SAMN30108760 VV21 VV21 mouse metagenome 1441287 SAMN30108761 VV23 VV23 mouse metagenome 1441287 SAMN30108762 VV25 VV25 mouse metagenome 1441287 SAMN30108763 VV27 VV27 mouse metagenome 1441287.
All procedures were approved by the Institutional Animal Care and Use Committees of University of Massachusetts Chan School and Wellesley College and performed in accordance with National Institutes of Health Animal Care and Use Guidelines.
Each author has made substantial contributions to the work: Conceptualization; formal analysis; writing—original draft; supervision, review and editing: KA, DW, MG, WV, BM, JK and MT. Methodology, review and editing: KA, RF, DW, MG, LT, DZ, and XH. Formal Analysis: KA, RF, DW, and MG, Project administration and funding acquisition: KA, WV, BM, JK, and MT. All authors have read and agreed to the published version of the manuscript.
This work was funded in part by NIH 5U24DK076169-13 Subaward # 30835-64 (KDA), SIAM Gravitation Grant 024.002.002 of the Netherlands Organization for Scientific Research (WMdV), NIH DK125407 and DK109677 (BAM), NIH 5U2C-DK093000 (JKK), and NIH DK61935 and Wellesley College Jenkins Distinguished Chair in Neuroscience Funds (MJT). Part of this study was performed at the National Mouse Metabolic Phenotyping Center (MMPC) at University of Massachusetts Chan Medical School.
WV is co-founder and has stock in The Akkermansia Company, and BM is a coinventor on a patent application PGT/US 18/42116 emanating, in part, from the findings described herein, and along with her respective academic institution, stands to gain financially through potential commercialization outcomes resulting from activities associated with the licensing of that intellectual property. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647078 | Lulu Wang,Yan Xiong,Beibei Fu,Dong Guo,Mohamed Y. Zaky,Xiaoyuan Lin,Haibo Wu | MicroRNAs as immune regulators and biomarkers in tuberculosis | 27-10-2022 | microRNA,Mycobacterium tuberculosis,host immunity,immune regulators,biomarkers | Tuberculosis (TB), which is caused by Mycobacterium tuberculosis (Mtb), is one of the most lethal infectious disease worldwide, and it greatly affects human health. Some diagnostic and therapeutic methods are available to effectively prevent and treat TB; however, only a few systematic studies have described the roles of microRNAs (miRNAs) in TB. Combining multiple clinical datasets and previous studies on Mtb and miRNAs, we state that pathogens can exploit interactions between miRNAs and other biomolecules to avoid host mechanisms of immune-mediated clearance and survive in host cells for a long time. During the interaction between Mtb and host cells, miRNA expression levels are altered, resulting in the changes in the miRNA-mediated regulation of host cell metabolism, inflammatory responses, apoptosis, and autophagy. In addition, differential miRNA expression can be used to distinguish healthy individuals, patients with TB, and patients with latent TB. This review summarizes the roles of miRNAs in immune regulation and their application as biomarkers in TB. These findings could provide new opportunities for the diagnosis and treatment of TB. | MicroRNAs as immune regulators and biomarkers in tuberculosis
Tuberculosis (TB), which is caused by Mycobacterium tuberculosis (Mtb), is one of the most lethal infectious disease worldwide, and it greatly affects human health. Some diagnostic and therapeutic methods are available to effectively prevent and treat TB; however, only a few systematic studies have described the roles of microRNAs (miRNAs) in TB. Combining multiple clinical datasets and previous studies on Mtb and miRNAs, we state that pathogens can exploit interactions between miRNAs and other biomolecules to avoid host mechanisms of immune-mediated clearance and survive in host cells for a long time. During the interaction between Mtb and host cells, miRNA expression levels are altered, resulting in the changes in the miRNA-mediated regulation of host cell metabolism, inflammatory responses, apoptosis, and autophagy. In addition, differential miRNA expression can be used to distinguish healthy individuals, patients with TB, and patients with latent TB. This review summarizes the roles of miRNAs in immune regulation and their application as biomarkers in TB. These findings could provide new opportunities for the diagnosis and treatment of TB.
Tuberculosis (TB), which is the 13th leading cause of death worldwide, is an infectious disease that seriously threatens human health. According to the World Health Organization Global Tuberculosis Report 2021, 1.5 million people worldwide died of TB in 2020, making it the second most lethal infectious disease after coronavirus disease 2019 (COVID-19) (1). Despite the devastating effects of TB, TB research has been limited by the COVID-19 pandemic, and conflicts in Europe, Africa, and the Middle East have caused disruptions in essential TB services and increased TB-related deaths, making TB one of the most lethal infectious diseases in the world. Mycobacterium tuberculosis (Mtb) is the pathogen that causes TB. Mtb is an obligate aerobic bacterium that is characterized by positive acid-fast staining (2). Although it possesses fimbriae and microcapsules, Mtb does not form spores. In addition, its bacterial wall contains neither the teichoic acid of gram-positive bacteria nor the lipopolysaccharide of gram-negative bacteria (3). Furthermore, Mtb has approximately 4000 genes with high guanine and cytosine contents (4). During infection, Mtb causes inflammatory responses and immune-mediated damage to the host (5). The main pathogenic substances of Mtb include capsules, lipids, and proteins (6). Additionally, although 2-7 h are required to kill Mtb in sputum with direct sunlight, Mtb is very resistant to dry, cold, acidic, and alkaline conditions. Therefore, Mtb can invade all tissues and organs of the body, with the most common site of infection being the lungs. Macrophages are the primary host cells of Mtb and the primary immune cells that function to clear Mtb. Macrophages eliminate Mtb by inducing apoptosis, autophagy, and inflammatory responses, which are critical for innate immunity (7). However, due to the dynamic nature of Mtb, it can interact with host cells to induce conditions that are more conducive to its survival. For example, Mtb alters the expression of host microRNAs (miRNAs) to regulate the expression of immune-related genes and promote its long-term survival. Furthermore, Mtb not only inhibits the clearance mechanisms of the host but also remains in host cells, ultimately causing latent tuberculosis infection (LTBI) (6). It is also estimated that one-quarter of the global population has LTBI, and of these patients, 5%–10% are at risk of developing active TB. As a result, failure to effectively control LTBI threatens the achievement of “End TB” goals. Antibiotics are the gold standard treatment for TB. However, although antibiotics (isoniazid, rifampicin, and streptomycin) are effective in the treatment of TB (8), Mtb can acquire drug-resistant phenotypes. Drug-resistant TB is simultaneously resistant to multiple drugs, meaning that multiple drugs have no therapeutic effect on TB (9). Thus, the increasing number of patients with drug-resistant disease each year complicates TB treatment. Mtb also uses immune evasion mechanisms to remain latent in host cells for a long time. Notably, previous studies have shown that Mtb can survive inside cells by regulating the miRNA expression of host cells (10, 11). miRNAs are a class of small noncoding RNAs that are approximately 20-24 nucleotides in length. Although they do not encode proteins, they are notably involved in regulating gene expression at the posttranscriptional level (10). miRNAs also inhibit gene expression by targeting the translation of specific mRNAs (12). Additionally, miRNAs play regulatory roles in many important physiological processes, such as cell proliferation and differentiation, body metabolism, and host immunity (13, 14). Infection with certain pathogens affect miRNA expression in the host, and miRNAs may play instrumental roles in directing immune responses (15). Moreover, in Mtb-infected cells, miRNAs appear to be used by Mtb to modulate host immunity (16). It has also been reported that the specific expression patterns of miRNAs can be used as potential diagnostic biomarkers in TB (17). Therefore, this paper mainly discusses the regulatory roles of miRNAs in metabolism, inflammatory responses, autophagy, and apoptosis during Mtb infection. We also summarize advances in the use of miRNAs as biomarkers in TB and further discuss the promises and challenges associated with their use as biomarkers.
Glycolysis is a common process of glucose catabolism that occurs in almost all biological cells and can occur under both aerobic and anaerobic conditions (18). Glycolysis involves the transformation of glucose into pyruvate through several enzymatic steps, simultaneously yielding adenosine triphosphate (ATP) and nicotinamide adenine dinucleotide. Glycolysis is the most critical process of glucose metabolism, and it the first metabolic pathway to be elucidated (19). When Mtb infects a host, glycolysis is a major metabolic process that promotes inflammatory responses in immune cells (20). To date, most studies on the immune-metabolic effects of Mtb infection have been conducted in macrophages. Efferocytosis is the process by which macrophages engulf and eliminate apoptotic cells, and it is used by the host to control infection when macrophages are exposed to pathogens (21). After Mtb infection, uninfected macrophages can phagocytose infected macrophages through efferocytosis (22). Based on these studies, Mtb infections result in poor innate immunity of macrophages, contributing to pathogen survival (23). Several authors have also shown increased activation of metabolic pathways in Mtb-infected macrophages (24–26). Other studies have shown that continuous glycolysis supports the production of the proinflammatory cytokine interleukin (IL)-1β, which, in turn, regulates prostaglandin 2 to control Mtb infection (27). After pathogen invasion, macrophages change their metabolic profile from oxidative respiration to high-rate aerobic glycolysis through the tricarboxylic acid cycle. This immunometabolic shift supports the production of the proinflammatory cytokine interleukin (IL)-1β, which promotes proinflammatory and antibacterial responses (28). miR-21 regulates intracellular glycolysis and limits macrophage metabolic reprogramming during Mtb infection. In a previous study, measurements of the extracellular acidification rate suggested that anti-miR-21 can substantially improve the glycolysis and glycolytic capacity of rat cardiomyoblast cells (29). Another study has proven that miR-21 impairs antimycobacterial responses by targeting IL-12 and B-cell lymphoma/leukemia-2 (Bcl-2) (30). It has also been shown that Mtb inhibits phosphor-fructokinase, muscle (PFK-M) via miR-21 to limit glycolysis in host cells (31). Lung tissues of mice infected with Mtb upregulate pri-miR-21 30 days after infection and maintain high levels of pri-miR-21 for 53 days. When murine bone marrow-derived macrophages (BMDMs) are infected with Mtb, miR-21 is continuously upregulated for 72 h and targets PFK-M at the critical step of glycolysis to inhibit this process. For the host, interferon-γ (IFN-γ), which drives host defenses against Mtb, inhibits miR-21 expression, forcing an isoenzyme switch in the PFK complex and maintaining PFK-M expression after Mtb infection. Therefore, miR-21 targets PFK-M to control macrophage immunometabolic function (31).
Lipids form fundamental components of cell membranes and are necessary for many physiological functions, such as energy supply, signal transduction, and cell recognition (32). Lipid metabolism refers to the synthesis, decomposition, digestion, and absorption of lipids by various enzymes. It also involves the processing of lipids into substances that are needed to maintain activities related to biological homeostasis. Evidence has shown that abnormal lipid metabolism can cause changes in membrane composition and permeability, resulting in the occurrence and development of various diseases (33). Simultaneously, specific changes in lipid synthesis and metabolism occur during pathogen invasion and carcinogenesis, and these changes facilitate pathogen survival and various malignant behaviors (34). miRNAs have recently been identified as critical regulators of lipid metabolic cycles, regulating the enzymes involved in lipid metabolism at the posttranscriptional level (35). This finding indicates that miRNAs are involved in the occurrence and development of various diseases by regulating lipid metabolism. The regulation of lipid metabolism by miR-33 has been intensively studied. Human miR-33 is located in the introns of the sterol regulatory element binding protein (SREBP) gene. Its mature form can be classified as miR-33a and miR-33b (36). miR-33 and its passenger strand (miR-33*) can target the key enzymes involved in cholesterol efflux, fatty acid metabolism, and insulin signaling, such as ATP binding cassette subfamily A member 1 (ABCA1), carnitine O-octanoyltransferase, and insulin receptor substrate 2 (37). A previous study reported that ABCA1 and ATP-binding cassette transporter G1 are targets through which miR-33a facilitates the SREBP-2-mediated regulation of cholesterol levels, thereby preventing the further removal of cholesterol from cells (38). Previous studies have reported that silencing miR-33 expression increases plasma high density lipoprotein (HDL) levels, reducing cholesterol flow to apolipoprotein A1 or neonatal HDL (39, 40). Many investigations have also indicated that the regulation of lipid metabolism by miR-33 is associated with many diseases. For example, several studies have shown that abnormal cholesterol metabolism is associated with neurodegenerative conditions, such as Alzheimer’s disease (41, 42) and age-related macular degeneration (43), and miR-33 is instrumental in these pathological processes. Mtb interacts with the host through complex lipid components in its cell wall; these interactions regulate metabolism and immune responses, thus affecting the physiological processes of the host cell and Mtb itself. Furthermore, since host lipids are the main source of nutrition for Mtb, the host can affect the outcome of an infection by regulating lipid homeostasis (44). Previous studies have shown that human macrophages infected with Mtb are induced to form lipid droplets (45). Moreover, high expression of lipid sequestration- and metabolism-related genes is observed in human TB granulomas, suggesting that the development of TB is related to the dysregulation of host lipid metabolism (46). It has also been reported that Mtb can not only use host intracellular lipid droplets as nutrients but also respond to host immune mechanisms by controlling the lipid contents in its cell wall (46). After murine peritoneal macrophages and macrophages derived from a transformed human mononuclear cell line (THP-1) are infected with Mtb for 48 h, miR-33 and miR-33* expression is upregulated through nuclear factor NF-kappa-B (NF-κB)-dependent mechanisms. After their upregulation, miR-33 and miR-33* negatively regulate mitochondrial fatty acid oxidation and extend lipid storage in macrophages. Finally, Mtb uses these macrophage lipids as a source of nutrients for its growth and reproduction (25). Although the upregulation of miR-33 favors the intracellular survival of Mtb, the effect of miR-33 overexpression on Mtb survival is still unclear. Overexpression of miR-33 may be detrimental to lipid storage in macrophages or may activate other immune responses in the host, which are not beneficial to the survival of Mtb. The effect of miR-33 overexpression on Mtb survival needs further study.
Inflammation is a defense response to stimuli and a basic physiological process that maintains homeostasis (46). When the body is subjected to physical damage, harmful stimuli (chemicals), pathogen invasion, tissue necrosis, or other harmful conditions that disrupt tissue homeostasis, the body initiates the inflammatory response (47). The typical inflammatory response involves several processes: the production of inflammatory inducers and the sensors that detect them, the production of inflammatory mediators, and the channeling of inflammatory mediators to target tissues (48). Moreover, it has been reported that complexes of soluble factors interact with cells during inflammation, which is a host response that accounts for tissue damage (49). These responses lead to the main manifestations of inflammation: redness, swelling, heat, pain, and dysfunction (50). Toll-like receptors (TLRs), which are one group of pattern recognition receptors, bind to pathogen-associated molecular patterns and induce the activation of multiple proinflammatory factors through the NF-κB and mitogen-activated protein kinase-related signaling pathways, resulting in inflammatory responses and pathogen clearance. Tumor necrosis factor-alpha (TNF-α), IL-1, IL-6, and IL-8 are among the inflammatory factors that are produced after TLR activation, and these factors subsequently play major roles as inflammatory cytokines. Members of the IL-1 family are central mediators of innate immunity and inflammation; most IL-1 family members (IL-1α, IL-1β, IL-18, IL-33, IL-36α, IL-36β, and IL-36γ) have proinflammatory activities, whereas some (IL-37 and IL-38) have anti-inflammatory effects (51). Previous studies have reported that cytokines and receptors of the IL-1 family are potent factors that prime and amplify the immune response, affecting nearly all cells that are involved in the innate immune system (52, 53). IL-6 can be produced by various cells, including monocytes, macrophages, dendritic cells (DCs), T cells, and B cells (54). Specifically, IL-6 exerts several effects on both immune and nonimmune cells (55); it can drive B-cell precursors to become antibody-producing cells (56), facilitate the growth and differentiation of primitive bone marrow-derived cells and enhance the lysis function of natural killer (NK) cells (57). In contrast, TNF-α is another prominent inflammatory mediator of inflammatory responses (58) that can activate neutrophils and lymphocytes, increase the permeability of vascular endothelial cells, regulate metabolic activities of other tissues, and promote the synthesis/release of other cytokines (59). Consequently, the presence of many inflammatory cytokines makes the body’s immune response function properly. When an organism is infected with a pathogen, the following occur: pathogen invasion, pathogen colonization of host tissue, immune response induction, pathogen clearance, or tissue damage. Inflammation is a link between innate and acquired immunity, helping the body to further eliminate pathogenic microorganisms and stimulating the initiation of acquired immune responses. Notably, intracellular bacteria multiply in host cells to escape attack by phagocytes, complement components, and antibodies. The common target cells of intracellular bacteria are epithelial cells, endothelial cells, hepatocytes, and macrophages (60). As a result, these bacteria, such as Mtb, Listeria, and Mycobacterium leprae, are difficult to eliminate from the host due to their intracellular life cycles. Subsequently, a structure called a granuloma forms in the infected area of the host when the host immune function is overwhelmed by the pathogen, resulting in chronic infection (61). Nevertheless, a few other pathogens survive and lie dormant in granulomas. If the granuloma ruptures, the pathogen can be reactivated and begin to proliferate (62). A previous study reported that when infected with Mtb, DCs, macrophages, and CD4+ T cells produce TNF-α and IL-12 in large quantities (63). Thus, NK cells enhance the ability of macrophages to phagocytose and kill pathogens by producing IFN-γ (64). The production of proinflammatory factors, such as IFN-γ, IL-12, and TNF, is essential for controlling Mtb infection (65). It was previously reported that miRNAs also play critical roles in controlling Mtb infection by regulating the inflammatory response and cytokine signal activation (66). A summary of the miRNAs that modulate the inflammatory response during Mtb infection is shown in Table 1 . Among the 24 miRNAs listed, 17 are upregulated and 7 are downregulated during Mtb infection. In macrophages, which are the main host cell of Mtb, eight miRNAs (miR-26a, miR-29a-3p, miR-32-5p, miR-125-5p, miR-132-3p, miR-155, miR-203, and miR-1178) (67–74) are upregulated and six miRNAs (miR-149, let-7 family, miR-20b, miR-27a, miR-18b, and miR-142-3p) (83–87, 89) are downregulated. However, in other types of immune cells, nine miRNAs (miR-21-5p, miR-99b, miR-144*, miR-30, miR-206, miR-140, miR-29b-1*, miR-124, and miR-223) (30, 75–82) are upregulated, and one miRNA (miR-378d) (88) is downregulated. A previous study showed that the miR-155-dependent downregulation of Src homologous 2-inositol phosphatase-1 (SHIP-1) could play a role in the survival of Mtb in infected mouse macrophages (24 h postinfection). As a direct target of miR-155, SHIP1 downregulation promotes the activation of serine/threonine kinase AKT, which is beneficial for Mtb survival (90). In adaptive immunity, miR-155 enhances IFN-γ production by human CD8+ and CD4+ T cells by targeting cytokine signaling-1 (91). However, miR-155 plays a dual regulatory role. Although miR-155 maintains bacterial survival in the early stage of macrophage infection, it promotes IFN-γ production by T cells to control Mtb infection in later stages (72). It may be that miR-155 plays different roles in innate immunity and adaptive immunity. miR-21 regulates not only host glycolysis but also inflammatory responses. After infection of RAW264.7 and THP-1 cells with Mtb for 24 h, the expression of miR-21-5p increases dramatically. miR-21-5p directly targets Bcl-2 and TLR4 in Mtb-infected macrophages to reduce the secretion of the inflammatory cytokines TNF-α, IL-1β, and IL-6, thereby allowing Mtb to evade the host immune response (92). miR-125b has been reported to directly target the 3’-UTR of κB-RAS2, an inhibitor of NF-κB signaling, increasing its stability and reducing inflammatory responses in primary human macrophages (91). Accordingly, in peripheral blood mononuclear cells of TB patients, miR−125b plays a vital role in the development and progression of TB by reducing the IFN-γ, IL-6, TNF-α, and NF-κB levels by inhibiting the Raf1 proto−oncogene serine/threonine protein kinase (93). As a major regulator of the cellular oxidative stress response, miR-144 directly targets nuclear factor erythroid 2-related factor 2 to modulate the oxidative stress response (94). The levels of miR-144*, which is the passenger strand of miR-144, are significantly elevated in the blood of TB patients, and this molecule regulates cytokine production by T cells (95). A previous study reported that miR-144* might regulate anti-TB immune responses by blocking the production of TNF-α/IFN-γ and inhibiting the proliferation of human T cells (76). However, no study has elucidated the specific mechanism by which miR-144* inhibits T-cell proliferation. The effect of miR-144* on T cell proliferation requires further study. In addition, miR-223, which is a small noncoding RNA, has been shown to be upregulated in the lung parenchyma and blood of patients with TB (82). In addition, studies have suggested that IL-6, chemokine ligand 3, and chemokine ligand 2 are novel targets of miR-223 (82, 96). Another previous study identified miR-29 as a central inhibitor of IFN-γ (68). As previously reported, the expression of miR-29 and IFN-γ is negatively correlated with Mtb infection, and miR-29 is downregulated in IFN-γ-secreting T cells and NK cells (97, 98). Moreover, promoting miR-29 expression increases susceptibility to mycobacterium infection (80). miR-29 has been identified as a regulator that suppresses immune responses to intracellular pathogens. Although miR-27a is expressed at low levels in Mtb-infected THP-1 macrophages (99), a previous study reported that miR-27a reduces the levels of IFN-γ, IL-β, IL-6, and TNF-α in macrophages by targeting IL-1 receptor-activated kinase 4 (an important kinase in the immune response), inhibiting the immune response, and enhancing the survival rate of intracellular Mtb (86). Similarly, miR-18b is downregulated in Mtb-infected human and murine macrophage cell lines. Recent studies have confirmed that low expression of miR-18b promotes the expression of hypoxia-inducible factor 1α, induces the production of proinflammatory cytokines and reduces the viability of bacteria in host cells (87). The levels of miR-20b, a member of the miR-17 family, are decreased in the serum and macrophages of patients with TB (100). A previous study reported that the activation of the NLR family pyrin domain containing 3 (NLRP3) facilitates the maturation of IL-1β and IL-18, eventually enhancing innate immune defenses (101). Furthermore, downregulation of miR-20b and upregulation of NLRP3 are observed in the macrophages of TB patients. In a TB mouse model, miR-20b was shown to directly bind to the 3’-UTR of NLRP3 and negatively regulate its expression. In summary, the downregulation of miR-20b increases the expression of NLRP3 and activates the NLRP3/caspase-1/IL-1β pathway to inhibit Mtb survival in macrophages (85). Previous studies have also indicated that miR-142-3p can target and inhibit the expression of IL-6 (102). Additionally, although miR-142-3p is downregulated in the peripheral CD4+ T cells and macrophages of patients with TB (89, 103), miR-142-3p expression is negatively correlated with the production of the proinflammatory mediators IL-6, NF-κB, and TNF-α. Therefore, decreasing miR-142-3p expression could delay the survival of Mtb in macrophages (89, 104). After infection of THP-1 and RAW264.7 macrophages with Mtb for 24 h, the downregulation of miR-378d causes the increase in Rab10 expression, which in turn leads to the increased expression of TLR4 on the cell surface and activation of the NF-κB, interferon regulatory factor 3, and MAPK signaling pathways (88, 105). As a result, a decrease in miR-378d expression can promote the production of the cytokines IL-1β, IL-6, and TNF-α, which is conducive to the clearance of intracellular Mtb.
Autophagy is a biological process that degrades intracytoplasmic macromolecules and organelles in capsular vesicles (106). During autophagy, part or all of the cytoplasm, including its organelles, are enclosed in double-membrane vesicles, forming autophagic vacuoles or autophagosomes. Soon after these autophagosomes are formed, they become monolayers and then combine with lysosomes to form autophagolysosomes (107). In autophagolysosomes, substances are decomposed into amino acids and nucleotides by various enzymes, and these amino acids and nucleotides can enter the tricarboxylic acid cycle to generate small molecules and energy (108). Additionally, the process of autophagy mainly involves the following stages: nucleation, elongation, formation and maturation of autophagosomes followed by fusion of autophagosomes and lysosomes. All these stages involve many genes, such as Beclin 1, AMP-activated protein kinase (AMPK), mammalian target of rapamycin complex 1 and autophagy-related genes (ATG) (109–111). Dozens of ATG and their homologs have been identified. A previous study reported that the whole process of autophagy is regulated by different ATGs (109). Autophagy targets bacteria in the cytoplasm or vacuoles, and this selective type of autophagy can be called xenophagy (112). A previous study reported that microtubule-associated protein light chain 3 (LC3)-modified autophagosomes form around target bacteria, and the pathogens are degraded through this LC3-associated phagocytotic process by promoting lysosome fusion with phagosomes (113). Moreover, increasing evidence suggests that autophagy can eliminate pathogens, but these pathogens can use various strategies to avoid being killed and to escape from phagosomes. Hence, pathogens block phagosome maturation, allowing their long-term survival in phagosomes (114). For example, Shigella foestri can competitively bind to the cell surface virulence protein IcsA through the invasion protein IcsB and block the binding of IcsA and ATG5 to avoid autophagic degradation (115). Listeria monocytogenes can also escape phagosomes via a toxin that forms pores in the phagosome membrane, enter the cytoplasm, and use the cell surface protein actin assembly inducing protein ActA to recruit host actin to the bacterial surface to prevent its recognition by autophagic machinery (116). Similarly, it has been demonstrated that Mtb has developed several strategies to evade autophagy. Among these strategies, Mtb can survive in cells by regulating miRNA expression profiles to avoid immune attack (117). As previously reported, the microtubule-associated protein LC3-I binds to phosphatidylethanolamine via a ubiquitin-like reaction that requires Atg7 and Atg3 (E1- and E2-like enzymes) (118). Mtb also reduces the Atg3 protein content through miR-155, negatively regulating autophagy (119). Moreover, the silencing of miR-155 during Mtb infection rescues autophagosome formation (119). Another previous study demonstrated that miR-155 is highly expressed in Mtb-infected mouse macrophages and enhances autophagy by targeting Rheb to inhibit Mtb survival (120). In addition, among the ATGs, Atg4 is a protease that is involved in converting LC3-I to LC3-II (121). A previous study also suggested that miR-129-3p is a repressor that facilitates the survival of Mtb in macrophages by targeting Atg4b-mediated autophagy (122). Similarly, miR-144-3p represses Atg4a expression by targeting its 3’-UTR, hindering the activation of autophagy (123). Atg7 has been reported to be involved in autophagosome formation and vesicle progression (124), whereas Atg16L1 controls the extension of nascent autophagosome membranes (125), indicating that Atg7 and Atg16L1 play essential roles in autophagy. As a member of the miR-17 family (126), there is increasing evidence that miR-106a regulates autophagy by targeting unc-51 like autophagy activating kinase 1 (ULK1), Atg7, and Atg16L1 (127, 128). A related study also indicated that miR-106a acts as a negative regulator of autophagy during Mtb infections, downregulating the expression of autophagy proteins by targeting ULK1 Atg7 and Atg16L1, thus inhibiting the autophagic process in macrophages (129). miR-20a can also target Atg7 and Atg16L1 to regulate autophagy and promote Mtb survival (130). Atg5 is a key protein involved in the elongation of phagocytic membranes in autophagic vesicles, and it forms a constitutive complex with Atg12 (131). Atg12-Atg5 then further binds to Atg16L to form an Atg12-Atg5-Atg16L complex, which is located on the outer membrane of the autophagosome (132). However, miR-1958 reduces the expression of Atg5 by interacting with Atg5’s 3’-UTR, inhibiting autophagy and promoting the survival of intracellular Mtb (133).
In addition to targeting ATG to regulate intracellular autophagy, miRNAs can regulate autophagy by targeting other components. For example, intracellular Ca2+ signaling regulates many basic cellular processes (134), and increasing evidence suggests that Ca2+ is a secondary messenger that regulates intracellular autophagy (135, 136). It has also been reported that calcium channels mediate the influx of calcium ions into cells upon membrane polarization. Elevated miR-27a expression has been observed in Mtb-infected cells, infected animals, and patients with active TB. Moreover, miR-27a directly targets calcium voltage-gated channel auxiliary subunit alpha2delta 3 (a Ca2+ transporter in the endoplasmic reticulum), inhibits the endoplasmic reticulum (ER) Ca2+ signaling pathway to reduce autophagy, and facilitates the intracellular survival of Mtb (137). Evidence has also suggested that the 3p and 5p arms of miRNAs perform the same or opposite functions in regulating gene expression (138, 139); this phenomenon was also observed for miR-30a. Specifically, miR-30a-3p provides a survival advantage for invading Mtb by inhibiting the maturation of autophagosomes and the fusion of mature autophagosomes with lysosomes (140). However, the activation of miR-30a-5p enhances autophagy, ultimately decreasing the growth of intracellular mycobacteria (141). It has been demonstrated that Mtb infection leads to downregulation of miR-17, which is accompanied by the upregulation of its target myeloid cell leukemia sequence 1 (Mcl-1) and signal transducer and activator of transcription 3 (STAT3, a transcriptional activator of Mcl-1) (142). Overexpression of miR-17 decreases the phosphorylation of protein kinase C-δ (an activator of STAT3) and the expression of Mcl-1 and STAT3. This suggests that during Mtb infection, downregulation of miR-17 inhibits autophagy through the miR-17–PKC-δ–STAT3–Mcl-1 pathway (143). A previous study demonstrated that TNF superfamily member 12 (TWEAK) enhance the expression of ATG in myotubes (144), suggesting that TWEAK may be involved in the regulation of autophagy. TWEAK is upregulated by mycobacterium components (Ag85A and Ag85B), and upregulated TWEAK induces phagosome maturation and promotes autophagy, ultimately decreasing intracellular mycobacterium survival (144). Increased miR-889 expression is observed in TB patients, and miR-889 inhibits autophagy to maintain the survival of Mtb via the posttranscriptional inhibition of TWEAK expression (145). Previous studies have also reported that UV radiation resistance associated gene (UVRAG) is involved in autophagy maturation and transport of endocytic vesicles to accelerate endocytic degradation (146, 147). UVRAG is crucial in the induction of autophagy, and it is the direct target of miR-125a-3p. After infection of mouse macrophages (RAW264.7 cells and BMDMs) with Mtb for 24 h, the increased expression of miR-125a-3p inhibits autophagy activation and antimicrobial effects against Mtb by targeting UVRAG (148). miR-125a-5p enhances autophagy by targeting the inhibition of STAT3 expression and blocks the intracellular survival of Mtb (149). In mouse macrophages, Mtb infection increases the expression of miR-23a-5p in a time- and dose-dependent manner (150). It has been reported that miR-23a-5p interacts with the 3’-UTR of Toll-like receptor 2 to inhibit its expression, impairing the TLR2/myeloid differentiation primary response gene 88 (MyD88)/NF-κB pathway and promoting the survival of Mtb (150). miR-18a belongs to the miR-17 family (151), and its expression gradually increases within 24 h after infection of RAW264.7 cells with Mtb. miR-18a directly targets and downregulates ataxia telangiectasia mutated (ATM) to inhibit autophagy and promote mycobacterial survival in macrophages. Furthermore, inhibition of miR-18a upregulates p-AMPK expression, which can be reversed by downregulating ATM. Therefore, the increased expression of miR-18a inhibits autophagy through the ATM-AMPK pathway, ultimately promoting intracellular Mtb survival (152). Similarly, DNA damage regulated autophagy modulator 2 (DRAM2) is a transmembrane lysosomal protein that is associated with autophagy processes (153), and it can interact with UVRAG to induce autophagy. miR-144* is expressed at notably high levels in Mtb-infected cells and interacts with the 3’-UTR of DRAM2 to reduce DRAM2 expression and autophagosome formation. As a result, miR-144* can decrease the antimicrobial response to Mtb by targeting DRAM2 (154). A previous study proved that miR-125b-5p can also target DRAM2 to inhibit antimicrobial responses in macrophages (155). Figure 1 shows the miRNAs involved in the regulation of autophagy after Mtb infection.
Apoptosis involves the activation, expression, and regulation of a series of genes. Apoptosis includes four stages: reception of apoptotic signals, interaction between apoptosis-regulating molecules, activation of proteolytic enzymes, and continuous reaction (156). Many studies have proven that apoptosis is induced by specific signals and that multiple genes coregulate apoptosis (157). For instance, the caspase gene family, p53 gene, Bcl-2 gene family, cellular myelocytomatosis viral oncogene, Fas cell-surface death receptor (Fas), and Fas ligand can all trigger apoptosis. Apoptosis is a physiological mechanism that maintains homeostasis. Some pathogenic factors inhibit or enhance apoptosis by targeting apoptosis-related genes, disrupting cell homeostasis and eventually causing various diseases (156). However, more apoptosis is not necessarily better. Excessive apoptosis exacerbates the outcomes of many diseases, such as neurodegenerative diseases, AIDS, and cardiovascular diseases (158). Insufficient apoptosis also results in disease. From the perspective of apoptosis, the pathogenesis of autoimmune diseases is caused by insufficient apoptosis and ineffective clearance of autoimmune T cells (159). Additionally, apoptosis is crucial for the elimination of infected cells from the host. Activation of apoptosis can effectively remove infected cells and terminate infection. However, the induction of apoptosis does not always protect host cells from microbial infection. Viruses and bacteria can exploit the host’s apoptotic machinery to reduce the number of cells that are needed for an immune response, allowing intracellular pathogens to escape clearance mechanisms and survive (160). There is a close relationship between viral infection and apoptosis (161). Viruses can cause tissue damage by increasing apoptosis rates (162, 163). There are also some viruses (such as poxviruses, herpesviruses, and adenoviruses) whose genomes encode antiapoptotic proteins that facilitate the completion of the viral replication cycle before apoptosis (164, 165), thereby ensuring viral replication and reproduction (166). In 1992, scientists discovered that bacteria could cause the apoptosis of infected host cells, so research on apoptosis also began to involve the field of bacterial infection (167). Similar to viral infections, bacterial infections promote or inhibit apoptosis. Different bacteria can affect the regulation of apoptosis. A single type of bacteria with different degrees of virulence may also differently affect apoptosis. Apoptosis has long been recognized to be an effective defense against the spread of mycobacterial infection (168). Mtb species with different degrees of virulence exert opposite effects on the apoptosis of macrophages. Attenuated mycobacteria induce apoptosis, and the growth of Mtb in macrophages is reduced, but Mtb is latent in macrophages (169, 170). Virulent Mtb blocks macrophage apoptosis, thereby maintaining its replicative niche and eventually causing host cell necroptosis to facilitate its escape and spread (171). Mtb utilizes multiple mechanisms to regulate host apoptosis. For example, the type VII secretion system ESX-1 secretion-associated protein EspC, a substrate protein secreted by Mtb, is thought to induce ER stress-mediated apoptosis (172). The type VII secretion system ESX-1 transcriptional regulator EspR is a DNA binding protein of Mtb that inhibits macrophage apoptosis through MyD88/TLR, providing opportunities for mycobacterial survival (173). The Mtb virulence factor phosphotyrosine protein phosphatase also promotes the intracellular survival of Mtb by inhibiting apoptosis (174). However, some proteins that are secreted by Mtb induce apoptosis (175–177).
Figure 2 shows how miRNAs regulate apoptosis in host cells after Mtb infection. miR-155 is not only an inflammatory regulator that performs dual functions but also plays dual regulatory roles in apoptosis. On the one hand, miR-155 promotes apoptosis to release Mtb antigens and activate T-cell immune function. On the other hand, it defends against apoptosis, allowing pathogens to escape and spread (178). miR-155 is upregulated in RAW264.7 macrophages after 12 h of M. bovis BCG infection. Furthermore, it has been shown that miR-155 enhances cAMP dependent protein kinase (PKA) signaling pathway activation by directly targeting protein kinase inhibitor alpha, which is a negative regulator of PKA signaling in macrophages. This process provides the main signal that drives macrophage apoptosis, resulting in loss of macrophage viability and favoring Mtb proliferation (179). Another study pointed out that miR-155 can inhibit apoptosis. miR-155 is upregulated in peripheral blood mononuclear cells (PBMCs) of patients with active TB, and it binds to the 3’-UTR of Forkhead box O3 (FOXO3) to inhibit its expression. It has been reported that the numbers of PBMCs in patients with active TB increase as a result of the inhibition of apoptosis by miR-155 (180). Similarly, miR-223 is upregulated in the monocyte-derived macrophages (MDMs) from patients with active TB and in infected THP-1 cells (181). miR-223 can also target FOXO3 to inhibit apoptosis (182). miR-20a-5p and miR-20b-5p are two highly homologous miRNAs, both belonging to the miR-17 family (183), but they perform different functions in regulating the apoptosis of Mtb-infected macrophages. The expression of miR-20a-5p is reduced in infected THP-1 macrophages, which is followed by c-Jun N-terminal kinase 2 (JNK2) and Bim activation. Mechanistically, miR-20a-5p directly targets JNK2 to regulate Bim expression, promoting apoptosis and Mtb clearance (184). Interestingly, miR-20b-5p downregulation causes effects that are diametrically opposed to those of miR-20a-5p. miR-20b-5p targets and negatively regulates Mcl-1, which increases cell viability and attenuates apoptosis in Mtb-infected macrophages (100). It has been demonstrated that miR-125b-5p is highly expressed in Mtb-infected macrophages and monocytes from TB patients. miR-125b-5p can target DRAM2 to decrease the expression of the apoptotic genes Bax and Bim, thereby inhibiting apoptosis (155). miR-21 acts as an apoptosis repressor in various tumor cells (185), and its antiapoptotic function is performed by upregulating the antiapoptotic factor Bcl-2 (186, 187). miR-21 also increases Bcl-2 expression in mouse monocyte macrophages (J774 macrophages) treated with the Mtb-derived protein Mpt64 (188). In BCG-infected mouse bone marrow-derived dendritic cells, miR-21 expression is increased, which inhibits Bcl-2, resulting in increased apoptosis (30, 92). miR-1281 can target and inhibit cyclophilin D, thereby protecting Mtb-infected human macrophages from programmed necrosis and apoptosis (189). The ligand of numb protein X 1 is an E3 ubiquitin ligase of NIMA related kinase 6 (NEK6) and a direct target of miR-325-3p. In Mtb-infected host cells, miR-325-3p is upregulated and blocks NEK6 degradation. Accumulation of NEK6 activates the antiapoptotic signal transducer and activator of transcription 3 signaling pathway, thereby promoting intracellular survival and the immune escape of Mtb (190). FOXO1 is considered to be a tumor suppressor and plays a proapoptotic role in various cells (191). miR-582-5p is abundantly expressed in the monocytes of patients with active TB and suppresses monocyte apoptosis by downregulating FOXO1 (192).
Biomarkers refer to biochemical molecules that indicate changes in a system, organ, tissue, cell, and subcellular structure or function. Biomarkers can help diagnose disease, determine disease staging, and evaluate the safety and efficacy of new drugs or treatments in target populations. Biomarkers not only include biological macromolecules, such as proteins and nucleic acids, but also include proteomes and metabolomes (193). They can be obtained from blood, urine, saliva, cancer cells, and cancer tissue samples. According to their functions, biomarkers can be classified into six categories: biomarkers for diagnosis, prognosis, prediction, efficacy, safety, and monitoring. Changes in host miRNA expression have diagnostic potential in TB (194). A summary of the differences in miRNA expression between patients with active TB and healthy individuals is shown in Figure 3 . Forty-three upregulated miRNAs and twenty-eight downregulated miRNAs were identified in different tissues of patients with active TB. Twenty upregulated miRNAs were identified in the serum of patients with active TB, of which 15 (miR-361-5p, miR-889, miR-576-3p, miR-16, miR-483-5p, miR-212, miR-220b, miR-650, miR-346, miR-125b, miR-378a-3p, miR-423-5p, miR-1249, miR-1178, and miR-668) are only upregulated in the serum (195–201). In patients with active TB, miR-22 expression is increased in the serum, but miR-22-3p is decreased in the plasma (197, 202). The miR-29 family members miR-29a and miR-29c are both upregulated in the serum and sputum of patients with active TB (197, 203). High levels of miR-20b are found in the serum and exosomes of patients, but opposite miR-20a expression patterns are found in serum (198, 204). Similarly, miR-146a expression in various tissue samples also follows diametrically opposing trends, and it increases in the serum and decreases in PBMCs (200, 205). Three miRNAs (miR-103a-3p, miR-107, and miR-148a-3p) are elevated only in the plasma of patients with active TB (202). The exosomal levels of miR-484, miR-425, miR-96, miR-486, and miR-185-5p increase in patients with active TB (204, 206, 207). For patients with active TB, the content of miR-191 is increased in exosomes and reduced in neutrophils (204, 208). Compared to unaffected healthy controls, miR-3179 and miR-147 levels in sputum (209), miR-589-5p and miR-199b-5p levels in PBMCs (210, 211), miR-331 and miR-204 levels in neutrophils (208), and miR-132 and miR-26a levels in MDMs all increased (67). miR-582-5p is increased in monocytes and PBMCs from patients with active TB (192, 211). miR-320 has unique expression patterns in patients; miR-320a is upregulated in neutrophils but downregulated in plasma, and miR-320b is downregulated in serum (197, 202, 208). miR-144* and miR-625-3p expression levels are increased in the T cells and urine of patients with active TB, respectively (76, 212). Three miRNAs (miR-155, miR-101, and miR-17) were only reduced in patient serum (196, 197, 199). The miR-30 family includes miRNAs that are downregulated in patients with active TB. For example, the expression of miR-30b and miR-30d in the serum and the expression of miR-30c in PBMCs are decreased (199, 213). In patients with active TB, seven miRNAs (miR-769-5p, miR-151a-3p, miR-223-3p, miR-448, miR-224-5p, miR-324-5p, and miR-488-5p) exhibited decreased levels only in the plasma (202, 214, 215), four miRNAs (let-7a-5p, miR-196b-5p, miR-892b, and miR-409-5p) exhibited decreased levels only in PBMCs (210, 211, 216), and two miRNAs (miR-197-3p and miR-99b-5p) exhibited decreased levels only in neutrophils (208). miR-365 is downregulated in the serum, monocytes, and macrophages of patients with active TB (217). Among the miRNAs summarized above, there are five miRNAs (miR-22, miR-20, miR-146a, miR-191, and miR-320) with distinctive expression patterns in different tissue samples, and these miRNAs may be useful as biomarkers to distinguish patients with active TB from healthy individuals. miRNAs can be used as markers for the early diagnosis of TB. miRNAs can be used not only to identify TB patients but also to monitor treatment effects, drug resistance, and Mtb virulence in TB patients ( Table 2 ). After anti-TB treatment, the expression of most miRNAs continued to be downregulated and gradually returned to the levels observed in uninfected healthy controls. The expression of seven miRNAs (miR-29a-3p, miR-155-5p, miR-361-5p, miR-99b, miR-29a, miR-146a, and miR-26) decreased in the plasma of treated patients (218, 219). Five downregulated miRNAs (miR-326, miR-346, miR-21-5p, miR-92a-3p, and miR-148b-3p) were identified in serum (199, 220, 221). Studies have shown that the miR-199b-3p, miR-199a-3p, and miR-16-5p levels in whole blood and the miR-424 levels in PBMCs are significantly decreased (222, 223). Increased miR-125a-5p levels were observed in TB patients who were treated for 2 months (221). As a unique miRNA in patients after treatment, miR-125a-5p expression can be combined with the greatly attenuated expression of other miRNAs to distinguish TB patients before and after treatment. Moreover, measuring miRNA expression can also Ibe used to determine the prognosis of TB patients. Antibiotics are an effective means of treating TB. However, due to the abuse of antibiotics or the insufficient course of treatment used by patients, common TB has developed drug resistance. For example, Mtb has acquired streptomycin resistance mechanisms (228). Drug resistance can be divided into monodrug, multidrug, and extensive multidrug resistance (229). The emergence of drug-resistant bacteria has greatly affected the treatment of TB. Therefore, the accurate diagnosis of drug-resistant TB and effective treatment are key factors in blocking the spread of drug-resistant Mtb. This article summarizes seven miRNAs that exhibit differential expression in patients with common and drug-resistant TB. The miR-4433b-5p and miR-424-5p levels in serum and the let7e-5p levels in exosomes are decreased in pan-susceptible TB patients compared to drug-resistant TB patients (224, 225). The miR-320a levels in the plasma and the contents of miR-197-3p and miR-223-3p in exosomes of drug-resistant patients are reduced (202, 225). miRNAs can be used to identify drug-resistant patients in order to select appropriate treatment regimens as soon as possible and design chemotherapy regimens according to the patient’s medication history, the prevalence of drug-resistant strains, and the available drugs. However, the use of miRNAs to diagnose drug resistance is limited, and changes in miRNA expression patterns caused by different drug-resistant strains and differences among drug-resistant patients also vary. The identification of broad-spectrum miRNAs that are applicable to all drug-resistant strains will help in the diagnosis of drug-resistant individuals, and the identification of specific miRNAs that can be used as biomarkers of different drug-resistant strains will be helpful for developing targeted therapies. Distinctive miRNA expression patterns are observed after infection with virulent (Mtb H37Rv), avirulent Mtb (Mtb H37Ra), and nonvirulent vaccine strains (M. bovis BCG). miR-145 expression in macrophages and miR-4484 expression in THP-1 cells is decreased after infection with the virulent strain (226, 227). Host cells upregulate six miRNAs (miR-125b, miR-4668-5p, miR-30e, miR-1275, miR-30a, and miR-3178) to inhibit the effects of the virulent strain (70, 227). The identification of differentially expressed miRNAs between cells infected with virulent and avirulent Mtb can be used to rapidly screen virulent strains to determine appropriate treatments. Changes in miRNA expression that are induced by Mtb infection may also be caused by other diseases in the body. Therefore, it is necessary to identify miRNAs that distinguish TB from other diseases, improving the accuracy of using miRNAs as biomarkers in the diagnosis of TB. miRNAs in the serum can distinguish patients with lung cancer, TB, and pneumonia. miR-21 and miR-155 are notably increased in the serum of patients with lung cancer and pneumonia compared to normal controls, and miR-182 is only crucially elevated in the serum of patients with lung cancer (230). One study measured plasma miRNA levels in patients with chronic obstructive pulmonary disease (COPD), asthma, and pulmonary TB. miR-21 and miR-34a are increased in patients with COPD and asthma, whereas miR-206 is decreased. miR-133 decreases in patients with COPD and TB and can be used to distinguish these patients from those with asthma (231). Crohn’s disease (CD) and intestinal TB have similar features and insensitive diagnostic tools, which makes their identification extremely difficult, but the use of miRNAs as biomarkers can solve this problem. The plasma miR-375-3p concentration is higher in patients with ITB than in patients with CD, whereas higher miR-375-3p expression is observed in the tissues of patients with CD (232). The clinical treatment and prognosis of tuberculosis pleural effusion (TPE) and malignant pleural effusion (MPE) are completely different, and effective biomarkers can quickly diagnose patients and enable them to receive effective treatment. miR-195-5p, miR-182-5p, and miR-34a-5p expression levels are much higher in patients with MPE than in patients with TPE, and they may be potential biomarkers for MPE diagnosis (233).
Mtb infection does not mean people will get sick. LTBI refers to host infection with Mtb without TB symptoms. When the host is infected with Mtb, the body’s immune cells cannot clear the pathogen in a timely manner. However, due to immune-mediated control of the infection, the patient does not experience the clinical symptoms of TB (6). Studies have shown that people with LTBI have a 5%-15% risk of developing active TB (234). Existing screening methods, such as interferon-γ release assays and tuberculin skin tests, are mainly used to diagnose Mtb infection based on a response of the patient to Mtb antigenic stimulation. However, these tests cannot differentiate active TB from LTBI, and both tests have lower accuracy in immunocompromised patients (235). Because no test has been found that effectively distinguishes LTBI from active TB, it is necessary to identify differentially expressed biomarkers between the two conditions. This review summarizes 39 miRNAs related to LTBI and TB ( Figure 4 ). Based on miRNA upregulation and downregulation in LTBI and TB, 39 miRNAs are divided into eight categories. Ten miRNAs (miR-1246, miR-2110, miR-370-3p, miR-28-3p, miR-193b-5p, let-7b-5p, miR-30b-5p, miR-424, miR-365, and 199a-3p) are upregulated only in TB patients (223, 236–238). Five miRNAs (miR-374a, miR-26a, miR-142-3p, miR-16-5p, and miR-451a) are downregulated only in TB patients (103, 239, 240). miR-424-5p and miR-27a-3p expression is decreased in patients with LTBI (241). Elevated expression levels of five miRNAs (miR-450a-5p, miR-140-5p, miR-21, miR-377-5p, and miR-3680-5p) are observed in patients with LTBI (236, 239, 241). The miRNAs that are mentioned above and only upregulated or downregulated in LTBI and TB patients cannot be used as biomarkers alone. More attention should be given to miRNAs that are at the intersection of expression in patients with LTBI and TB. miR-150 and miR-146a expression is decreased in patients with TB and LTBI (222, 239). Of course, there are miRNAs (miR-29a-3p, miR-361-5p, miR-196b, miR-451a, miR-340-5p, miR-199b-3p, miR-6856-3p, miR-16-5p, miR-374c-5p, miR-6886-3p, and miR-378) whose expression is increased in both patients with TB and LTBI (218, 223, 242–244). The number of miRNAs with elevated expression in LTBI and TB is higher, which may be caused by the host immune response to pathogen infection. In addition, four miRNAs are differentially expressed in patients with TB and LTBI. let-7e-5p, let-7d-5p, and miR-212-3p are upregulated in patients with LTBI but downregulated in patients with TB (236, 245). miR-376c is higher in patients with TB, but it is decreased in patients with LTBI (242).
As a new type of biomarker, miRNAs have promising application prospects (1). miRNAs are very stable. As short-chain noncoding RNA molecules with only 22 nucleotides, miRNAs are extremely stable in serum, plasma, urine, and other samples (246, 247). miRNAs maintain good stability even after prolonged storage and freeze−thaw cycles (248). This property is advantageous for the use of miRNAs as potential biomarkers for clinical disease (2). A wide range of miRNA sources are available. miRNAs can be easily obtained from cells, extracellular fluids, and body fluids. Changes in the expression of miRNAs can be rapidly measured by quantitative polymerase chain reaction (3). miRNAs can serve as biomarkers of TB in multiple ways. Commonly used TB detection methods can effectively identify infected patients. However, ideal results from other perspectives, such as distinguishing LTBI from TB, identifying drug-resistant patients, and determining the prognosis after treatment, cannot be obtained. miRNAs can distinguish TB patients from healthy controls, TB patients from LTBI patients, treated patients from untreated patients, and TB patients from patients with other diseases. miRNAs can even help to identify the drug resistance and virulence of Mtb (4). In this study, six miRNAs (miR-29, miR-361, miR-146, miR-26, miR-199, and miR-16) were identified that can distinguish between healthy individuals and TB, TB and LTBI patients before and after treatment. miR-212, miR-378, miR-196, miR-365, and miR-30 are differentially expressed in healthy individuals, TB patients, and LTBI patients, and miR-30 can also identify virulent strains. miR-148, miR-346, miR-155, miR-99, and miR-125 can be used as biomarkers to distinguish healthy individuals, TB patients, and TB patients after treatment, and miR-125 can also identify virulent strains. miR-223, miR-197, let-7, and miR-320 can not only indicate infected patients but also identify drug-resistant patients (5). When searching PubMed with the keywords “micro-RNA” and “Biomarker,” nearly 30,000 results were found. miRNAs not only can be used for the diagnosis of TB but also have excellent application prospects in aging-related diseases and cancer. The use of miRNAs as biomarkers also faces many challenges. First, many miRNAs that can be used as biomarkers have identified, but the mechanisms of most miRNAs remain unclear. Second, there are differences in miRNA expression patterns between individuals. Similarly, differences in the age and sex of subjects causes different results in the use of miRNAs as biomarkers. In addition, the strength of the immune system can also affect the results. Suitable miRNA biomarkers should be stable enough to be useful in this role under dissimilar conditions. Finally, miRNAs should be disease specific, and reliable biomarkers should be able to differentiate TB, LTBI, and other respiratory diseases. To solve these problems, the following solutions can be adopted in future research. First, miRNAs that have the potential to be used as biomarkers should be examined, and the specific mechanisms by which these miRNAs function should be thoroughly explored. Next, the combination of multiple miRNAs can mitigate the differences that are observed between individuals and improve the accuracy of the detection results. Finally, standardized miRNA expression data can be established to differentiate TB, LTBI, and other respiratory diseases.
TB, which is a highly contagious disease, is difficult to control or completely cure with traditional treatment and detection methods in a timely manner. It is important to develop effective tests and treatments for TB. Accumulating evidence suggests that miRNAs are of considerable importance for host immunity. Some miRNAs positively regulate the immune response to clear pathogens in response to Mtb infection. However, Mtb can upregulate the expression of certain miRNAs and suppress the expression of immune-related genes to evade clearance mechanisms. This suggests that we can target miRNAs to enhance the function of the immune system to treat TB. In addition, we still need to pay attention to some limitations in the current research. First, a large number of studies have focused on how miRNAs regulate inflammation, autophagy, and apoptosis while ignoring the important role of miRNAs in regulating metabolism. Mtb continuously exchanges substances and energy within host cells to maintain its own growth and reproduction. The identification of miRNAs that specifically inhibit the metabolic pathway of Mtb may be possible to fundamentally cure tuberculosis. Second, although many miRNAs have been shown to target certain genes to regulate anti-Mtb immune responses, their specific molecular mechanisms remain unclear. Can one microRNA target multiple genes and can these genes interact with each other during Mtb infection? Could this interaction play a greater role in amplifying antibacterial effects or would they counteract each other? Finally, certain miRNAs may play opposite roles in innate and acquired immune responses. At present, studies on the roles of miRNAs in antibacterial immunity have mainly focused on innate or acquired immunity, but these branches of the immune response function as a whole and should not be separated. For some miRNAs with potential antibacterial effects, the functions and specific molecular mechanisms of their target genes should be explored in representative innate and acquired immune cells. At present, there are some tests that can detect tuberculosis quickly and effectively. However, these traditional methods are less effective in distinguishing among TB patients, LTBI patients and healthy individuals. LTBI has a strong incubation period, and if not detected and treated with preventive drugs, such patients will develop active TB. Compared with traditional methods, detecting differentially expressed miRNAs in patients is more efficient and rapid. Only real-time fluorescence quantitative polymerase chain reaction is required to obtain the results. In addition, the combination of multiple miRNAs can improve the accuracy of detection results. As treatment progresses, miRNA expression patterns will sensitively change, which is beneficial for doctors when administering targeted treatment to patients. However, there are no systematic and standardized quantitative methods for measuring miRNA expression for clinical diagnosis. This greatly affects the accuracy and application of miRNAs as biomarkers for TB diagnosis. Therefore, in the future, miRNAs can be used as targets to advance research on the treatment and diagnosis of TB in order to cure this disease that plagues humans as soon as possible.
LW, XL, and HW drafted the initial manuscript with feedback from all authors. YX, BF, DG and MZ all gave their comments and suggestions to the manuscript. XL and HW reviewed and modified the manuscript. All authors read and approved the final manuscript.
This work was supported by the National Natural Science Foundation of China (No. 81970008, 82000020), the Fundamental Research Funds for the Central Universities (No. 2022CDJXY-004, 2019CDYGZD009 and 2020CDJYGRH-1005), Natural Science Foundation of Chongqing, China (cstc2020jcyj-msxmX0460) and Chongqing Talents: Exceptional Young Talents Project (No. cstc2021ycjh-bgzxm0099). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
The authors would like to thank Dr. Tao Li (Wuhan University) and Dr. Rui Wang (Sun Yat-sen University) for providing helpful comments and critical suggestions.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647083 | Mei-Ling Yang,Richard G. Kibbey,Mark J. Mamula | Biomarkers of autoimmunity and beta cell metabolism in type 1 diabetes | 27-10-2022 | type 1 diabetes,glucose metabolism,posttranslational modifications,biomarkers,neoepitopes | Posttranslational protein modifications (PTMs) are an inherent response to physiological changes causing altered protein structure and potentially modulating important biological functions of the modified protein. Besides cellular metabolic pathways that may be dictated by PTMs, the subtle change of proteins also may provoke immune attack in numerous autoimmune diseases. Type 1 diabetes (T1D) is a chronic autoimmune disease destroying insulin-producing beta cells within the pancreatic islets, a result of tissue inflammation to specific autoantigens. This review summarizes how PTMs arise and the potential pathological consequence of PTMs, with particular focus on specific autoimmunity to pancreatic beta cells and cellular metabolic dysfunction in T1D. Moreover, we review PTM-associated biomarkers in the prediction, diagnosis and in monitoring disease activity in T1D. Finally, we will discuss potential preventive and therapeutic approaches of targeting PTMs in repairing or restoring normal metabolic pathways in pancreatic islets. | Biomarkers of autoimmunity and beta cell metabolism in type 1 diabetes
Posttranslational protein modifications (PTMs) are an inherent response to physiological changes causing altered protein structure and potentially modulating important biological functions of the modified protein. Besides cellular metabolic pathways that may be dictated by PTMs, the subtle change of proteins also may provoke immune attack in numerous autoimmune diseases. Type 1 diabetes (T1D) is a chronic autoimmune disease destroying insulin-producing beta cells within the pancreatic islets, a result of tissue inflammation to specific autoantigens. This review summarizes how PTMs arise and the potential pathological consequence of PTMs, with particular focus on specific autoimmunity to pancreatic beta cells and cellular metabolic dysfunction in T1D. Moreover, we review PTM-associated biomarkers in the prediction, diagnosis and in monitoring disease activity in T1D. Finally, we will discuss potential preventive and therapeutic approaches of targeting PTMs in repairing or restoring normal metabolic pathways in pancreatic islets.
Posttranslational modifications (PTMs) change the properties of a protein and shape its biological functions (1). Various pathways altered by PTMs that arise from tissue inflammation have been closely linked to numerous disorders including cancer and autoimmune diseases (2, 3). Type 1 diabetes (T1D) is a chronic autoimmune disease characterized by altered glucose sensing and insulin response resulting that may arise from the immune attack of insulin-secreting beta cells in the pancreas. In addition, tissue specific properties of pancreatic islets, and beta cells in particular, may contribute to the autoimmune pathology (4, 5). However, we are reminded that T1D, similar to other autoimmune syndromes, are multifactorial in origin, including a role for genetics, stochastic factors, and environmental influences in the onset and progression of disease. It is clear that some, or many, tissue specific PTMs may not be expressed in the thymus in the course of immune tolerance induction, though clear studies of specific PTMs are lacking in fully understanding central tolerance to modified proteins. However, T cells specific to PTM determinants can escape selection from the immune system, providing the potential for the modified proteins to be recognized as neo-antigens and contribute to autoimmunity. As one clear example, the role of citrullination PTM has been extensively studied in rheumatoid arthritis (RA). As with T1D, RA is also a chronic autoimmune disease characterized by inflammation of the target tissue, connective tissue in the joints, and more than 100 citrullinated proteins have been identified from RA synovium (6). Anti-citrullinated protein antibodies (ACPAs), present before the early onset of RA and correlate with disease severity, are routinely used for the diagnosis of RA for over a decade (7, 8). Similarly, citrullination has recently become a relevant PTM in T1D pathology (9). Accumulating evidence has identified significant numbers of citrullinated islet proteins, including proteins in the glucose and insulin metabolic pathways. These citrulline PTM proteins elicit vigorous B and T cell autoimmune responses in both human T1D and NOD murine disease, including glutamic acid decarboxylase 65 (GAD 65), 78-kDa glucose-regulated protein (GRP78) (also called BiP, HSP5a), islet antigen-2 (IA2), islet-specific glucose 6 phosphatase catalytic subunit-related protein (IGRP), islet amyloid polypeptide (IAPP) and glucokinase (10). Indeed, several other inflammatory PTMs also play the vital roles in the progression of T1D, including deamidation, oxidation and carbonylation (11). Not surprisingly, several enzymes responsible for forming, repairing and/or regulating PTMs such as peptidylarginine deiminase (PADs), antioxidant enzymes, catalase, glutathione peroxidase 1 (GPx1), and superoxide dismutase (SOD) are also found to modulate T1D autoimmunity and glucose and insulin metabolism. However, it is important to realize that it is not clearly known if specific PTMs are a cause of pathology or a consequence of pathology in human T1D. Animal models and ex-vivo studies have been partly useful in defining the roles of PTMs in disease but do not perfectly reflect human disease. Beta cells have very specific intracellular pathways that are involved in coupling the metabolism of glucose to the release of insulin. Many components of these pathways are subject to modification by PTMs, including citrullination. The details of how beta cell metabolism is coupled to insulin secretion through oscillatory activation of the phosphoenolpyruvate (PEP) cycle to close KATP channels have been recently reviewed in depth (12) and will only be summarized here. Glucose enters the rodent beta cell through glucose transporter 2 (Glut2) and is introduced into glycolytic metabolism relative to its concentration in the plasma by the activity of glucokinase (GK). GK is not product inhibited unlike the other hexokinases and has an EC50 for glucose in the physiologic range. The glucose carbons then flow through glycolysis and enter the mitochondria as pyruvate and follows one of two pathway fates. The first pathway is the more familiar pyruvate dehydrogenase pathway (where pyruvate is converted to acetyl CoA), which is ultimately oxidized to CO2 by the TCA cycle and electron transport chain in the process of oxidative phosphorylation (OXPHOS). OXPHOS supports the basal ATP requirements of the cell. As the ATP/ADP ratio reaches its thermodynamic equilibrium the OXPHOS progressively slows in the process known as ADP privation where the mitochondrial membrane potential hyperpolarizes, and TCA cycle intermediate accumulate (13). In particular, acetyl CoA increases in response to the high matrix NADH/NAD+ and mitochondrial GTP production is increased driven by the high ATP/ADP ratio (via antiparallel collaboration of the ATP and GTP isoforms of succinyl CoA synthesis) (14). Increased acetyl CoA activates pyruvate carboxylase diverting pyruvate into anaplerotic synthesis of oxaloacetate (OAA) consuming an ATP in the process. In the mitochondrial matrix, OAA is converted into phosphoenolpyruvate (PEP) coupled to GTP hydrolysis by the mitochondrial isoform of PEPCK (PCK2) (15). Thus, the newly made, highly energetic PEP exits the matrix where it is used to raise the ATP/ADP ratio via hydrolysis back to pyruvate by pyruvate kinase (PK). The cycle from pyruvate to OAA to PEP and back to pyruvate is known as the PEP cycle. Because PK raises the ATP/ADP ratio more than the mitochondria bioenergetics permit and because it is physically localized to the KATP channel, it closes KATP channels to depolarize the plasma membrane to allow Ca2+ to enter to stimulate insulin release (13). The cellular workload from membrane depolarization and insulin release increases ATP hydrolysis resetting the system and returning control over ATP synthesis back to OXPHOS. Subsequent cycles of such coordinated metabolic and electrical oscillations are partially entrained by the changes in cellular work as well as the generation of fructose-1, 6-bisphosphate (F16BP) by the PFK1/PFKFB3 system. F16BP allosterically activates PK and supports mitochondrial ADP privation and KATP closure. Many of the components of this glucose-sensing mechanism in beta cell are modified by PTMs triggered by beta cell stress ( Table 1 ). Relevant to metabolic pathways, we will summarize the conditions that elicit PTMs, including their role in immunological and biological processes, with specific focus on their implications in T1D. We will particularly highlight the inflammatory PTMs triggered by beta cell stress, the effects of immunometabolic targets/biomarkers on T1D autoimmunity and beta cell metabolism and discuss PTMs-based approaches for preventive and therapeutic options of T1D.
One pathologic hallmark of juvenile onset T1D is the immense lymphocyte infiltration around and within pancreas islets, i.e., inflammation termed as insulitis. Insulitis includes the massive liberation of proinflammatory cytokines and reactive oxygen species (ROS), a microenvironment that enhances a wide variety of resident protein PTMs. Both experimental and clinical studies demonstrate that beta cell stress-induced PTMs participate in neo-antigen formation, beta cell dysfunction and even beta cell death in the initiation and progression of T1D ( Figure 1 ) (11, 56). Essentially all professional secretory cells rely on endoplasmic reticulum (ER) functions, a protein folding factory, to ensure that accurately folded synthesized proteins find their way to the secretory pathway. However, the process of protein folding is often altered by various cell stresses including viral infection, chemical exposure, heat shock, ROS and inflammation. Improperly folded proteins accumulate in the ER to disrupt ER homeostasis. The ER has finely mechanisms to govern protein quality control by the unfolding protein response (UPR), ER-associated degradation (ERAD) and autophagy (57). One beta cell can synthesize more than 3000 insulin molecules per second (58). Both misfolded proinsulin monomers and aggregates during insulin biosynthesis are primarily cleared by ERAD pathway (59). Therefore, it is not surprising that secretory pathways of beta cell are susceptible to ER stress caused by inflammation and autoimmunity (60–62). For example, ORP150 is an ER resident HSP 70 family chaperone induced by ER stress. Autoantibody against to ORP150 is detected in patients with T1D (63). GRP78 acts as a sentinel to inactivate the UPR pathway by inhibiting ER stress sensor membrane proteins, including protein kinase RNA (PKR)-like ER kinase (PERK), activating transcription factor 6 (ATF6), and inositol-requiring protein 1 (IRE1). While it is not clear that citrullination alters GRP78 activity, it is also found in synovial fluid and antibody against citrullinated GRP78 frequently detected in patients with RA (6, 64). Similarly, citrullinated GRP78 was found in human islets under cytokine-induced stress in vitro and antibody against citrullinated GRP78 was also detected in patients with T1D (37). In support of PTMs that drive autoreactive inflammatory processes, there is higher frequency of circulating CD4+ T cells against citrullinated GRP78 peptides in T1D PBMC compared to healthy subjects. However, it is clear that the peripheral T cell compartment may not accurately reflect tissue resident T cell populations. Deletion of the IRE1-X-box–binding protein 1 (XBP1) pathway in pancreatic beta cells results in decreased oxidative folding of proinsulin and insulin secretion along with decreased expression of protein disulfide isomerases (PDIs) (65, 66). Over 30% of proteins require PDI as a chaperone to catalyze disulfide bond formation and facilitate protein folding including preproinsulin, proinsulin and insulin (67). PDIA1, also called prolyl-4-hydroxylase beta (P4Hb), is highly expressed in pancreatic islets and are required for proinsulin oxidative folding in vitro (34, 68, 69). Endoplasmic reticulum oxidase 1 (ERO1), another abundant protein expressed in the pancreatic islet, is responsible for recycling reduced PDIA1/P4Hb by FAD cofactor for transferring electrons to oxygen. ERO1-β mutant mice develop impaired glucose-stimulated insulin secretion and decreased insulin content in islets (70). Deficiency of ERO1-β increased cell apoptosis in MIN6 beta cells treated with tunicamycin, an inhibitor of n-glycosylation, resulting in protein misfolding and ER stress (71). Calcium is essential for the activity of many ER-resident chaperones to ensure accurate protein folding. Sarco/endoplasmic-reticulum calcium ATPase (SERCA) is responsible to pump calcium from cytosol into the ER lumen to maintain higher ER intraluminal Ca2+ levels (100-800 µM compared to 100 nM Ca2+ in the cytosol) (72). Carbonylation, another PTM amplified in inflamed tissues, leads to a loss of sarco(endo)plasmic reticulum Ca2+-ATPase (SERCA2a) activity and diastolic dysfunction in the streptozotocin (STZ)-induced T1D murine model (73). Frequently, ER stress results in Ca2+ leakage from ER lumen and then activates Ca2+ -dependent PTM enzymes such as tissue transglutaminase 2 (Tgase2) and PADs. Marré et al. reported that chemically-mediated ER stress induced immunogenicity of murine CD4+ diabetogenic BDC2.5 T cells mediated by increased Tgase2 activity (74). Recently, Donnelly et al. found that Tgase modified-GAD65 and -IA2 increased the binding affinity of these PTM ligands to their corresponding serum autoantibodies from patients with T1D (75). Cytosolic PAD enzyme catalyzes the irreversible deimination to convert arginine into citrulline within proteins, a pathway that is closely regulated by calcium. Under the physiologic Ca2+ concentration, PADs maintain normal basal activity. When cytosolic calcium concentration increased to 100-fold higher (approximately 1-100µM) above normal physiological concentration in response to cell stresses such as inflammation and ER stress, PAD enzymes become fully activated (76, 77). Among five PAD isozymes, PAD2 has the highest mRNA and protein expression level and activity in the pancreatic islets from C57Bl/6, non-obese diabetes resistance (NOR) and NOD mice (38). However, there is no PAD2 mRNA expression in C57Bl/6, NOR and NOD liver, another major organ for maintaining glucose homeostasis outside of the pancreas (78). Of note, PAD2 and PAD4 are the only PAD isozymes expressed in immune cells and their corresponding enzyme activity in synovial fluid positively correlates with RA tissue inflammation and disease activity (79). Moreover, a pan-PAD inhibitor, BB-Cl-amidine was found to prevent diabetes in the NOD murine model (80). Recently, we have carefully reviewed the role of PAD enzymes in the pathogenesis of T1D development (9). PADs require reducing conditions for efficient catalytic activity. For example, PAD enzyme in the synovial fluid from patients with RA catalyzes citrullination of human fibrinogen in vitro in the presence of reducing agents, dithiothreitol (DTT) or reduced glutathione (GSH) (81). Of note, GSH is the most abundant endogenous antioxidant. In addition, the level of ROS also regulates PAD enzyme activity. Damgaard et al. reported that H2O2 inhibited the catalytic activity of recombinant human PAD2 and PAD4 in vitro (82). Recently, Kim et al. reported that H2O2 promoted cellular senescence mediated by the inhibition of PAD2 expression in osteoblasts (83). However, how PAD enzyme is regulated and how aberrant PADs activity leads to pathogenic conditions are still not clear. For example, several studies demonstrated that patients with RA have higher oxidative stress and lower GSH level compared to healthy subjects (84–86). Recently, Nagar et al. reported that thioredoxin, the other major redox regulator besides GSH, can activate the enzyme activity of PADs (87). Their study provides one of the mechanisms why citrullination level is increased in patients with RA while the level of GSH, the known co-activator of PADs, is decreased. Collectively, these studies indicate that PAD enzyme activity and citrullination levels in individual tissues and tissue proteins are susceptible to oxidative stress and redox imbalance. Of note, oxidative stress induced by hyperglycemia and insulitis plays a key role in the onset of T1D and diabetes-related complications of disease. Elevated biomarkers of oxidative stress are frequently detected in tissue, urine and blood from patients with metabolic disorders including T1D and T2D (88–90). It has been hypothesized that beta cells express lower levels of antioxidant enzymes compared to other tissues. The expression of catalase, GPx and both cytosolic Cu/Zn SOD and mitochondrial Mn SOD in mouse islets are lower compared to liver, kidney, brain, heart, lung, skeletal muscle, heart muscle, adrenal gland, and pituitary gland (91, 92). In addition, the expression of catalase and GPx is reduced in human pancreatic beta cells compared to alpha cells. Moreover, beta cell viability is reduced after oxidative stress in H2O2 or NO treated human islets (93). Consistent with these observations, serum levels of GPx and SOD are reduced in patients with T1D compared to healthy subjects (94). Thus, antioxidants may possess anti-diabetic potential in NOD murine model (95, 96) supporting this therapeutic strategy in patients with T1D (97, 98). Insulin and its precursors, preproinsulin and proinsulin, also undergo PTM including oxidation and deamidation (39–41). Notably, insulin A-chain (A1-13) with a vicinal disulfide bond between A6-A7 was required for T cell recognition by using a CD4+ T cell clone isolated from an HLA DR4+ child with autoantibody against insulin (39). Several common PTMs, including citrullination, deamidation, chlorination and oxidation, increase HLA-A*02:01- binding affinity to insulin-B-derived epitopes in vitro (45). The deamidation of glutamine catalyzed by Tgase2 is also found to modulate T cell recognition to beta cell autoantigens. Van Lummel et al. reported that deamidation increases epitope binding affinity to HLA DQ by using Tgase2-modified peptides including phogrin, IA-2, IGRP, GAD65 and proinsulin (41). Moreover, there are autoreactive CD4+ T cells against deamidated insulin B30-C13 found in patients with early onset T1D. Protein carbonylation, the major PTM product of oxidative stress, contributes to insulin resistance and metabolic dysfunction in adipose tissue of both animal models and human T1D (99). Hyperglycemia induces oxidative stress, the major stress to trigger and amplify carbonyl modification, and then leads to pancreatic beta cell and endothelial cell dysfunction (100). In adipose tissue, oxidative stress induced GLT4 carbonylation and resulted in GLT4 activity loss (101). Other studies have clearly profiled carbonylated plasma proteins as potential biomarkers in T2D (102–104). Of note, Telci et al. reported that the level of plasma carbonyl PTMs were increased in adolescent and young adult T1D patients compared to the healthy subjects (105). Carbonylated pancreatic amylase and chymotrypsinogen were identified as biomarkers for autoimmune pancreatitis and fulminant T1D, respectively (106, 107). Our recent study also defined a group of pancreatic beta cell proteins with carbonylation, all bound by autoantibodies from human and NOD mice T1D antisera including PDIA1/P4Hb, PDIA2, 14-3-3 protein isoforms, GRP78 and chymotrypsinogen B (35). Of interest, carbonylated PDIA1/P4Hb was found to be an early autoantigen, triggering both autoantibodies and autoreactive T cells in human T1D.
T1D is a T cell mediated autoimmune disease that both CD4+ and CD8+ T cells are involved in the selective attack of insulin producing beta cells. Emerging data demonstrate that T cell responses are finely linked to bio-energetic metabolism, glycolysis and oxidative phosphorylation (OXPHOS). In a quiescent state, T cells favor the use mitochondrial OXPHOS to make ATP for basal energy production. Upon antigen stimulation, T cells are reprogrammed to use glycolysis to adjust the increased energy demand to support cell activation, proliferation and differentiation. Inhibition of glycolysis to manipulate T cell autoimmunity has been tested in several autoimmune diseases including SLE, MS, RA and T1D (20, 108, 109). Some of these pathways will be further defined below. The major biological function of pancreatic beta cells is to secret insulin in response to the change of glucose concentration to maintain glucose homeostasis. In the development of T1D, inflammation and oxidative stress amplifies various PTMs within islet self-proteins which then break immune tolerance in addition to altering beta cell metabolism ( Figure 2 ) ( Table 1 ).
Glucose transporter 1 (GLUT1) facilitates the metabolic switch to glycolysis in activated T cells. 2-Deoxy-D-glucose (2DG), a glucose analog, is taken up by GLUT and then converted to 2DG-6-phosphate by hexokinase in cytoplasm where it is no longer metabolized. 2-DG-6-phosphate accumulating in the cells inhibits hexokinase and phosphoglucose isomerase to then block glycolysis. In comparison to quiescent T cells, activated T cells are more susceptible to 2-DG due to the upregulated GLUT1 expression and glycolytic metabolism. Treatment of NOD mice with 2DG results in the reduction of diabetogenic CD8+ T cells specific to IGRP (NRP-V7 epitope), less lymphocyte infiltration within the islets and improves beta cell granularity (16). GLUT1 blockade therapeutic strategy is also considered for T1D patients that undergo islet transplantation to potentially protect beta cell loss due to graft rejection (110). Glucose uptake is mediated by GLUT2 in rodent beta cells (111). However, it remains controversial if GLUT 1, 2 or 3 are individually critical for glucose uptake in human beta cells (112). Relevant to glucose uptake, SGLT2 (sodium-glucose co-transporter-2) facilitates renal glucose reabsorption from the circulation. SGLT2 inhibitors reduce renal glucose uptake threshold and have been utilized in patients with T2D to lower plasma glucose levels, with limited risk of hypoglycemia and to prevent cardiovascular complications (17). Since both elevated urine glucose and ROS levels increase SGLT2 activity, several preclinical and clinical studies are ongoing to evaluate the antioxidant effects of SGLT2 inhibitors mainly for patients with T2D but also in streptozotocin (STZ)-induced diabetic murine models (113). Recently, Shyr et al. reported that SGLT2 inhibitors protect from glucotoxicity-induced beta cell failure through mitigation of oxidative and ER stress (114). Of note, several studies evaluated the efficacy and safety of SGLT2 inhibitors in patients with T1D (112, 115).
Glucokinase, mainly expressed in the liver and pancreatic beta cells, is the first rate-limiting step of glycolysis in glucose metabolism. However, the metabolic roles of glucokinase in liver and pancreas are fundamentally different for glycogen synthesis and insulin secretion, respectively. Glucokinase (hexokinase IV) belongs to the family of hexokinases. Unlike other hexokinase I-III, glucokinase activity is not regulated by feedback inhibition by its product, glucose-6-phosphate. Glucokinase has ~35-fold lower affinity for glucose (S0.5 7-9 mM) compared to other hexokinases (S0.5 ~0.2 mM). In addition, the small fluctuations of its enzyme activity alter the threshold of glucose-stimulated insulin secretion in pancreatic β-cells. Therefore, glucokinase is believed to act as an important glucose sensor by controlling the rate of glucose input into in pancreatic beta cell metabolism. More than 600 mutations of human glucokinase gene have been identified in patients with glucokinase linked hyperinsulinemic hypoglycemia (PHHI-GK), glucokinase-linked permanent neonatal diabetes (PDNM-GK) and glucokinase-linked maturity-onset diabetes of the young (MODY-GK, also called MODY-2). Several studies also demonstrate that glucokinase activity is regulated by PTMs. For example, polyubiquitination of human glucokinase, both pancreatic isoform 1 and hepatic isoform 2, allosteric activates glucokinase catalytic activity up to 1.4 fold (18). SUMOylation (small ubiquitin-like modifiers) of glucokinase was found in MIN6 and INS-1 murine cell lines and results in increased pancreatic glucokinase stability and activity (19). The STZ-induced diabetic mouse model of T1D exhibits decreased glucokinase expression with hyperglycemia (116). Recently, we found that citrullination decreases the catalytic activity and substrate binding affinity of human pancreatic glucokinase and diminishes glucose stimulated insulin secretion (GSIS) in INS-1E murine cells (10). In addition, citrullinated glucokinase is present in NOD pancreas prior to insulitis and in human islet beta cells exposed to inflammatory cytokines. Moreover, immune self-tolerance is broken by citrullination as indicated by the presence of autoantibodies and autoreactive CD4 T cells against to citrullinated glucokinase in patients with T1D. Glucose sensing and proliferative capacity differs significantly between immature and mature beta cells, though both secrete insulin. Immature beta cells sense lower glucose concentrations via hexokinase 1 and gradually lose proliferative functions as they mature (117). A major difference of metabolic machinery is in the switch of expression from high glucose affinity hexokinase 1 in immature beta cells to low glucose affinity glucokinase (also known as hexokinase 4) in mature beta cells (118). We recently demonstrated that citrullination increases the Km of glucokinase by 2-fold (10). This seemingly small change in Km is nonetheless significant since glucokinase functions in a narrow range of glucose concentration near the Km. While PTMs may trigger increased turnover of modified proteins, there is no known degradation pathway yet identified for protein citrullination. Thus, even small molar changes in irreversible glucokinase citrullination may reflect long term abnormalities in glucose sensing and insulin secretion in individual islets. As defined throughout this review, PTMs that arise may separately alter metabolic processes and/or trigger specific autoimmune responses, and the two outcomes may separately alter or contribute to pathology in the pancreas.
Besides GLUT1, inhibition of the glycolysis pathway enzymes to modulate T cell autoimmune responses may be an attractive therapeutic strategy for T1D. For example, a small molecule PFK15, a competitive inhibitor of 6-phosphofructo-2-kinase/fructose-2, 6- biphosphatase 3 (PFKFB3), is found to inhibit glycolysis and T cell response to beta cell antigens in diabetogenic CD4+ T cells from NOD.BDC2.5.TCR.Tg mice (20). In addition, PFK15 treatment delayed diabetic onset in the adoptive transfer model of T1D by BDC2.5 CD4+ T cells. Interestingly, PFKFB3 expression is upregulated in beta cells from patients with pre-T1D and T1D compared to non-diabetic subjects (21). Given that several PTMs regulate the biological activity, proteosomal degradation and stability of PFKFB3 in cancer cells (119), it may yet be an important target for therapeutic manipulation of diabetogenic T cells.
Enolase is the glycolytic enzyme that converts 2-phosphoglycerate (2PG) to phosphoenolpyruvate (PEP). The tissue distribution of α-enolase in various autoimmune syndromes has not yet been fully investigated with a potential role in immune mediated tissue pathology. For example, the expression of neuron-specific enolase (NSE) was not detected in the pancreas of autopsied T1D patients, but was present in the islets of non-diabetic subjects (22). Interestingly, anti-α-enolase autoantibodies have been identified in numerous autoimmune diseases such as autoimmune retinopathy, SLE, RA, MS, IBD (120–122). Although PTM modifications of α-enolase with have not yet been reported in autoimmune diabetes, the presence of citrullinated α-enolase in the RA joint and autoantibodies against to citrullinated α-enolase in RA serum suggests that citrullinated α-enolase can initiate and drive chronic inflammatory responses in autoimmune diseases (123).
Pyruvate kinase is the last enzyme in glycolysis pathway and produces net ATP and pyruvate. The catalytic activity of PK is tightly regulated by PTMs including phosphorylation, acetylation, citrullination, methylation, succinylation and glycosylation (24–26) as well as through allosteric interaction with F16BP mentioned above. For example, PK activity is increased 2 to 3-fold after in vitro citrullination by PAD (124). Our recent studies demonstrate that one pan-PAD inhibitor, YW3-56, can restore cytokine-mediated suppression of insulin secretion upon pyruvate stimulation in INS-1E beta cells (10). The expression of PKM1, PKM2 and α-enolase is down regulated in renal glomeruli from patients with T1D compared to healthy control subjects (23). Pharmacologic activation of pyruvate kinase M2 protects against diabetic nephropathy by increasing glucose metabolic flux and inducing mitochondria biogenesis (27). In rodent and human islets, PK activation accelerates the frequency of metabolic oscillations and increases GSIS in vivo (13, 125). Chronic treatment with PK activator also protects islet function on a high fat diet. Upon TCR engagement, PKM2 will translocate into the nucleus of T cells. Treatment with TEPP-46, an allosteric activator of PKM2, blocks its nuclear translocation, inhibits Th1 and Th17 polarization mediated by glycolysis blockade in vitro and ameliorates the development of EAE murine model (126). This is another pathway in beta cell metabolism where PTM modification may alter PK biology, yet not affect autoimmune specific responses. There are no defined autoimmune responses to PK in T1D. Thus, PTMs may also singularly affect metabolic components without subsequent autoimmune specific responses (as is the case with PK). The alternative observation that PTMs trigger only autoimmune responses without affecting specific metabolic pathways has been defined in many studies for other self proteins.
Pyruvate is the final product of glycolysis. Once pyruvate enters into mitochondria, it can be metabolized either by pyruvate dehydrogenase (PDH) or pyruvate carboxylase (PC) to enter into TCA cycle metabolism. PDH allows pyruvate to enter into the oxidative TCA pathway through the generation of acetyl CoA in the mitochondrial matrix. PDH activity is regulated by PTMs such as phosphorylation, acetylation and succinylation (28, 29). Beta cell-specific PDH deficient (β-PDHKO) mouse strains develop increased blood glucose level and decreased plasma insulin level in the first month after birth presumably by decreasing glucose oxidation (30). In addition, GSIS was reduced in isolated islets from β-PDHKO mice compared to age-matched control mice. Of note, Zurgil et al. reported that anti-PDH autoantibodies were found in several autoimmune diseases including primary biliary cirrhosis, Sjogren’s syndrome, scleroderma, SLE and RA (127). Interestingly, double knock out the inhibitory PDH kinases in beta cells (that increase PDH activity) actually leads to decreased insulin secretion (125). Thus, the right balance of glucose oxidation versus anaplerosis is required for functional beta cell metabolism (125).
Comparison of pyruvate oxidation through PDH versus pyruvate carboxylation via PC demonstrates that flux through the latter strongly correlates with GSIS in INS1 cells, rodent and human islets (125, 128). While PC is most strongly regulated via allosteric activation by acetyl CoA levels (in addition to the direct measures of its flux noted above), its relevance to GSIS has been assessed by knockdown and overexpression approaches. GSIS was not reduced in pyruvate carboxylase siRNA treated INS-1 cells presumably because of incomplete knockdown, while it was suppressed when treated with the chemical inhibitor of PC, phenylacetate (129). In contrast, overexpression of PC in INS-1 cells increases insulin secretion and cell proliferation (31). PC expression is down regulated in islets from patients with T2D compared to non-diabetic subjects (130). To protect human beta cells from inflammation and nitrosative stress, pyruvate carboxylase (PC) is needed for promoting glutathione (GSH) synthesis and suppressing NO synthesis to limit ROS and NO level, respectively (32, 33).
P4Hb, a member of the PDI family and the beta subunit of a tetramer of prolyl-4-hydroxylase (P4H), is the most abundant ER oxidoreductase for the retention and accurate folding of proinsulin/insulin in pancreatic beta cells (131, 132). Ablation of the thioredoxin activity by chemical modification of PDIA1/P4Hb prevents the refolding of denatured and reduced proinsulin in vitro (34). Therefore, PDIA1/P4Hb plays the critical role of proinsulin processing, insulin secretion and protection from ER stress in islet beta cells. Recently, we found that carbonylated P4Hb is elevated in human islets under inflammatory and oxidative stress and is coincident with decreased glucose-stimulated insulin secretion and altered proinsulin to insulin ratios (35). We identified that carbonylated PDIA1/P4Hb serves as an antigenic islet protein supported with the presence of autoantibody and autoreactive T cells against carbonylated PDIA1/P4Hb in patients with T1D. In a small population of early onset T1D patients (n=21, under 1 year of disease duration), we found that 76% patients had either anti-PDIA1/P4Hb alone (11 out of 21 patients) or anti-PDIA1/P4Hb linked with anti-insulin (auto)antibodies (5 out of 21 patients). In contrast, no patient had anti-insulin IgG (auto)antibodies without the presence of anti-PDIA1/P4Hb antibodies, indicating a potential link between these two autoantibody subsets in T1D. Moreover, PDIA1/P4Hb plasma level were increased in pre-diabetic NOD mice and in children with T1D, newly diagnosed within 48 hours (36). In a small cohort of T1D patients (<14 yrs of age) followed longitudinally, some autoantibody responses to P4Hb appear transient, suggesting that antibodies may reflect acute stress in the pancreas. It suggests PDIA1/P4Hb as a potential immunometabolic biomarker for early diagnosis of T1D and also provides mechanistic insight of carbonylated P4Hb into insulin metabolism and neo-epitope formation in the progression of T1D. It is hypothesized that carbonyl modification of P4Hb may cause altered folding of insulin, causing both and accumulation of proinsulin and/or creating an immunogenic misfolded form of insulin itself.
While exogenous insulin therapy is still the central intervention to treat T1D, a more complete understanding of T1D as a heterogeneous disease with multiple affected immunologic and metabolic pathways encourages versatile modalities to treat, delay and even prevent T1D. Such strategies include immune modulation, islet specific strategies to prevent inflammation, and improved glycemic management. The successful clinical trial of Teplizumab, a FcR non-binding anti-CD3 mAb, reported a delay in the median time to diagnosis of 2 years compared to placebo group for relatives of patients with T1D (133, 134). Importantly, a new era of therapeutic strategy exists with T cell mediated therapy, specifically upon FDA approval of Teplizumab for at-risk T1D individuals to delay the onset of this debilitating disease. Besides anti-CD3 mAb therapy, many attempts of monoclonal antibody and antibody derivatives target on other ligands of T cells and B cells such as CD2, CD20, CD80, CD86 and cytokines such as TNF-α, IL-21, IL-2 and IL-6R are actively ongoing for T1D immune-focused therapy. Other strategies to reduce beta cell stress, maintain islet antigen immune tolerance, sustain glucose homeostasis and even combination therapy are also leading to more therapeutic options for T1D. There are several in-depth reviews recently summarized current and the future therapies for T1D (135–138). Herein, we highlight knowledge gap about potential PTM-based T1D therapy and PTM biomarkers that may reflect diagnosis, disease activity and/or assistance for establishing optimal timing of T1D treatment ( Figure 3 ). Strollo et al. reported that autoantibody against oxidative modified insulin (oxPTM-insulin) and insulin autoantibody (IAA) co-existed in 50% of patients with T1D (42). Of note, 34% of IAA negative T1D patients were oxPTM-insulin positive. In this study, the oxidative modification of insulin antigen includes chlorination of Tyr16 and Tyr26, oxidation of His5, Cys7 and Phe24, and glycation of Lys29 and Phe1 in chain B. Interestingly, some PTMs (citrullination, chlorination, deamidation, and oxidation) can increase the binding affinity of insulin-B-derived peptides on HLA-A*02:01 compared to their counterpart native peptides (45). Strollo et al. also tested the sensitivity and specificity in comparison of oxPTM-insulin antibody with other established T1D autoantibodies (43). Moreover, anti-oxPTM-insulin was observed to precede the onset of T1D in prediabetic children (44). Their studies suggest antibody against oxidative modified insulin as a potential biomarker for better diagnosis compared to current diagnostic T1D autoantibodies and even as a biomarker for prediction of T1D in children. The observations imply that PTM-insulin (oxidation) breaks immune tolerance, leading to autoreactive B and T lymphocytes. Thereafter, subsequent epitope spreading may result in the production of antibody against the non-PTM self insulin protein. As with autoantibody responses to P4Hb, as above, anti-oxPTM-insulin antibodies may be associated with the onset and/or susceptibility of diabetes-associated complications. The concept of epitope spreading upon breaking of immune tolerance by PTM self proteins has been supported in models of lupus autoimmunity (139–141). The Mamula laboratory demonstrated that elevated isoaspartyl PTM content is found in lupus-prone mice. Particularly, isoaspartyl PTM is associated to lupus T cell proliferative defect in MRL mice (142). Both snRNP D and histone H2B are known lupus autoantigens. Specifically, T cell immune tolerance is broken in isoaspartyl snRNP D immunized mice and subsequent to activate B cells producing antibodies against to both isoaspartyl PTM and naïve forms of snRNP D peptide (140). Moreover, autoantibodies that bind both isoaspartyl and aspartyl form of histone H2B are present in human SLE and lupus-prone MRL/lpr mice, yet another example of epitope spreading (141). It must be emphasized that the clear roles of most PTMs in human disease are not fully understood, notably, whether PTMs are a cause of pathology or a consequence on existing tissue pathology. Moreover, the bias of ex vivo technology in defining T cell specificity from peripheral cells may not accurately reflect tissue resident T cell populations and/or their role in pathology. In a similar manner as described above, protein carbonylation is the major product of oxidative stress. Recently, we reported that antibodies against carbonyl-PDIA1/P4Hb and native PDIA1/P4Hb often co-exist in patients with established T1D (35). Interestingly, antibody against PDIA1/P4Hb precedes the onset of hyperglycemia in murine NOD mice, as early as 4 weeks of age. In line with the promising anti-CD3 preventive T1D trial in children, PTM relevant autoantibody analysis (either anti-oxPTM-insulin and anti-PDIA1/P4Hb antibodies) may help establish the optimal time for therapeutic immune modulation of high-risk T1D individuals. Sultan et al. reported that high glucose and oxidative stress increased protein carbonylation and glutathione peroxidase (GPx) activity in HUVECs cells (143). The enhanced GPx activity is due to the compensation of decreased GPx1 protein expression in high glucose or methylglyoxal-treated HUVECs cells. The data suggest that Lys114 carbonylation of GPx1 may alter the substrate H2O2 biding affinity that increases catalytic activity. It is conceivable that antioxidant approaches to mitigate beta cell stress and/or reduce oxidative PTMs is a potential therapeutic strategy for T1D. Verapamil, a calcium channel blocker, rescued mice from STZ-induced hyperglycemia mediated by preventing beta cell loss (144). Moreover, verapamil decreased thioredoxin-interacting protein (TXNIP) expression in islets from STZ injected mice and TXNIP is one of the important redox regulators in cells. In a phase II clinical trial of 32 adults with recent onset T1D (diagnosed within 3 months), the verapamil treated group had improved glycemic control and mixed-meal-stimulated C-peptide secretion compared to placebo group (145). As noted earlier, calcium is essential for PAD activity. Verapamil fully inhibited Ca2+ influx in both A549 and THP-1 cells and fully blocked protein citrullination in A549 cells (146). The effect of verapamil in beta cells on modifying PAD, citrullination, ER stress and other cellular events remains to be more thoroughly investigated for evaluating the therapeutic potential of verapamil in T1D. Akin to PAD enzymatic activity, Tgase2 (also known as tissue transglutaminase) is a calcium-dependent enzyme that catalyzes deamidation reaction. Anti-Tgase2 antibody serves as a serological marker of celiac disease, thought to be mechanistically associated with T1D (147–149). In particular, Maglio et al. reported anti-Tgase2 antibody deposition in the small intestine of a majority of children with T1D (150). In addition, anti-Tgase2 with combination of IAA, anti-GAD65 and anti-IA2 is found to facilitate screening for pre-T1D and celiac disease (151). Recently, Tgase2 inhibitor, ZED1227, was reported to attenuate gluten-induced small intestinal damage compared to placebo group in a phase II clinical trial of 41 patients with celiac disease (152). However, the therapeutic potential of Tgase2 inhibitors for patients with T1D has not yet been fully investigated. Several citrulline blockade approaches have been developed and studied in RA. For example, several potent PAD4 specific reversible inhibitors can disrupt mouse and human NET formation (NETosis), thought to be a major source of autoantigens in RA (153–155). Recently, Sodre et al. reported that BB-Cl-amidine, a pan-PAD inhibitor, prevented diabetes development in the NOD murine model mediated by reduced pancreas citrullination level and autoantibody against citrullinated GRP78 (80). Recently, we demonstrated that a PAD2/PAD4 inhibitor, YW3-56, partially restored cytokine-mediated suppression of insulin secretion upon glucose or pyruvate stimulation in INS-1E cells and citrullination disrupted pancreas glucokinase activity (10). While inhibitors of PTMs may be a promising preventative therapeutic approach for T1D, the potential side-effects and/or risks of long term use in young children requires more thorough consideration, given that disease pathology may be chronic in development. Glucokinase activators have been evaluated in patients with T2D including piragliatin, MK-0941, AZD1656 and dorzagliatin (156, 157). Recently, a phase II clinical trial reported that TTP399, a novel hepatoselective glucokinase activator, lowered HbA1c and reduces hypoglycemia without increasing the risk of ketosis compared to placebo group (158). Last year, the FDA granted a Breakthrough Therapy designation for TTP399 as an adjunctive therapy to insulin for T1D patients based on the promising result by the above clinical trial. Relevant to citrullination, Rituximab, a B cell depleting anti-CD20 antibody, was reported to modulate ACPAs level in patients with RA compared to placebo group (159). In phase II clinical trials, Rituximab delayed the decline of c-peptide, i.e. preserved beta cell function, in patients with T1D compared to placebo group but the effect was transient (160, 161).
Herein, we have identified how specific PTMs may arise in T1D and their consequences on both autoimmunity and metabolic pathways, notably glucose sensing and insulin release. While there are examples of PTMs affecting both autoimmune specificity and metabolism, the effects of PTMs may also be clearly distinct and delineated. For example, a number of PTMs have been noted that trigger autoimmune response without obvious or defined alterations in beta cell metabolism. Conversely, specific PTMs may only affect metabolic pathways in the absence of autoimmune specificity. Thus, therapies that may ‘correct’ PTMs as they arise in T1D may have important consequences to various stages and distinct pathways that characterize T1D. Individual PTMs may be subject to therapeutic manipulation (reversal), while others are permanent and irreversible. Some specific PTMs serve as a common platform for coordinating metabolism and countering beta cell stresses arising from environmental factors (such as diet, virus and chemical) or microenvironmental factors (such as cytokines, ER stress and calcium fluctuation). Similarly, PTMs may serve as biomarkers that may predict and better diagnose early steps in T1D will prove important for designing therapies for preserving and rescuing beta cell functions.
M-LY, RK and MM conceived the concept and co-wrote the manuscript. All authors contributed to the article and approved the submitted version.
M-LY and MM were supported by the JDRF (3-SRA-2017-345-S-B, 1-SRA-2020-977-S-B and 1-INO-2022-1116-A-N) and RK by NIH DK127637.
The authors would like to acknowledge helpful discussion of concepts with colleagues, including Drs. Kevan Herold, Tukiet Lam, Lut Overbergh, Eddie James, Carmella Evans-Molina, Steven Clarke, Hubert Tse and Cate Speake.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647089 | Anna Rita Minafra,Alexandra Chadt,Puyan Rafii,Hadi Al-Hasani,Kristina Behnke,Jürgen Scheller | Interleukin 6 receptor is not directly involved in regulation of body weight in diet-induced obesity with and without physical exercise | 27-10-2022 | interleukin 6 (IL-6),obesity,exercise,diet,mouse model | High level of interleukin 6 (IL-6), released by adipocytes in an obesity-induced, low grade inflammation state, is a regulator of insulin resistance and glucose tolerance. IL-6 has also regenerative, anti-inflammatory and anti-diabetogenic functions, when secreted as myokine by skeletal muscles during physical exercise. IL-6 mainly activates cells via two different receptor constellations: classic and trans-signalling, in which IL-6 initially binds to membrane-bound receptor (IL-6R) or soluble IL-6 receptor (sIL-6R) before activating signal transducing gp130 receptor. Previously, we generated transgenic soluble IL-6 receptor +/+ (sIL-6R+/+) mice with a strategy that mimics ADAM10/17 hyperactivation, reflecting a situation in which only IL-6 trans-signalling is active, whereas classic signalling is completely abrogated. In this study, we metabolically phenotyped IL-6R deficient mice (IL-6R-KO), sIL-6R+/+ mice and wild-type littermates fed either a standard chow (SD) or a high-fat diet (HFD) in combination with a 6-weeks treadmill exercise protocol. All mice were subjected to analyses of body weight and body composition, determination of blood glucose and insulin level under fasting conditions, as well as determination of substrate preference by indirect calorimetry. Neither classic IL-6 nor trans-signalling do influence the outcome of diet-induced obesity, insulin sensitivity and glycaemic control. Furthermore, IL-6R deficiency is not impairing the beneficial effect of physical exercise. We conclude that the IL-6R does not play a requisite role in regulation of body weight and glucose metabolism in diet-induced obese mice. | Interleukin 6 receptor is not directly involved in regulation of body weight in diet-induced obesity with and without physical exercise
High level of interleukin 6 (IL-6), released by adipocytes in an obesity-induced, low grade inflammation state, is a regulator of insulin resistance and glucose tolerance. IL-6 has also regenerative, anti-inflammatory and anti-diabetogenic functions, when secreted as myokine by skeletal muscles during physical exercise. IL-6 mainly activates cells via two different receptor constellations: classic and trans-signalling, in which IL-6 initially binds to membrane-bound receptor (IL-6R) or soluble IL-6 receptor (sIL-6R) before activating signal transducing gp130 receptor. Previously, we generated transgenic soluble IL-6 receptor +/+ (sIL-6R+/+) mice with a strategy that mimics ADAM10/17 hyperactivation, reflecting a situation in which only IL-6 trans-signalling is active, whereas classic signalling is completely abrogated. In this study, we metabolically phenotyped IL-6R deficient mice (IL-6R-KO), sIL-6R+/+ mice and wild-type littermates fed either a standard chow (SD) or a high-fat diet (HFD) in combination with a 6-weeks treadmill exercise protocol. All mice were subjected to analyses of body weight and body composition, determination of blood glucose and insulin level under fasting conditions, as well as determination of substrate preference by indirect calorimetry. Neither classic IL-6 nor trans-signalling do influence the outcome of diet-induced obesity, insulin sensitivity and glycaemic control. Furthermore, IL-6R deficiency is not impairing the beneficial effect of physical exercise. We conclude that the IL-6R does not play a requisite role in regulation of body weight and glucose metabolism in diet-induced obese mice.
In the last two decades, obesity has been described not only as elevated amount of fat cells caused by excess of nutrients and a low degree of physical inactivity, but also associated with an inflammatory state. The transition from healthy lean to obese adipose tissue is accompanied by a chronic low-grade inflammation and immune dysregulation, as well as the enhanced release of pro-inflammatory cytokines, which can consequently interfere with peripheral insulin signalling and glucose metabolism. Among others, IL-6 has been frequently associated with the impaired immune control in obese adipose tissue (1). Among several studies, Wallenius et al. showed that mice lacking IL-6 developed mature-onset obesity, associated with a disturbed carbohydrate and lipid metabolism (2). These data were subsequently supported by Matthews et al., who observed increased body weight, impaired glucose tolerance and exacerbated insulin resistance in IL-6 KO mice (3). Moreover, enhanced inflammation in liver and skeletal muscles and insulin resistance was observed in hepatocyte-specific IL-6R deficient animals (4), as well as increased insulin resistance in whole-body IL-6 KO mice (5) and increased body weight in astrocyte-specific IL-6 deficient mice (6). In contrast, adipocyte-specific deletion of IL-6 in the context of diet-induced and genetic obesity had no effect on body weight and fat content, glucose tolerance and insulin resistance (7) or, in another study, it determined slightly reduced high fat diet (HFD)-induced glucose intolerance (8). Taken together, it is still unclear whether IL-6 is a primary trigger for the development of obesity and insulin resistance or whether it is actually required to counteract the increased inflammation associated with obesity. Additionally, this intricate scenario is complicated further by the observed beneficial effect of IL-6 produced by skeletal muscles following physical exercise. Indeed, during intense exercise, both IL-6 mRNA and protein levels increase in skeletal muscles (9, 10) and plasma IL-6 rises up to 100-fold (11). In addition, there are many lines of evidence that IL-6 has also regenerative effects, transiently downregulates immune function and can actually protect from obesity and insulin resistance. Indeed, during physical exercise, IL-6 promotes blood glucose disposal and blood glucose uptake in skeletal muscles by stimulating cell surface glucose transporter 4 (GLUT4) translocation in muscle cells (1). In addition, it may increase fatty acid uptake, lipolysis and free fatty acids release from adipocytes and skeletal muscles, respectively (12, 13). Moreover, IL-6 has been shown to stimulate pancreatic insulin production and insulin sensitivity in peripheral tissues, including skeletal muscle and adipose tissue, together with enhancing skeletal muscle hypertrophy and bone remodelling (14). Mechanistically, two main signalling pathways can be activated by IL-6. In the classic signalling pathway, IL-6 binds to its membrane-bound receptor (IL-6R), followed by dimerization of glycoprotein 130 (gp130) and activation of JAK/STAT, MAPK, and PI3K/AKT (15). In the trans-signalling pathway, IL-6 can bind soluble IL-6 receptor (sIL-6R) molecules which are generated via ectodomain shedding by metalloproteases (ADAM-10 and ADAM-17) (16) or through alternative splicing of IL-6R mRNA (17). Of note, activation of classic IL-6 signalling is limited to specific tissues, since IL-6R is only expressed in distinct cell types, such as immune cells and hepatocytes. Some studies suggest that IL-6R might be expressed also in adipocytes and myocytes (18–20). Thus, it is not yet clear whether metabolic functions of IL-6 mainly rely on classic or trans-signalling. Accordingly, here, we metabolically characterized the previously generated IL-6 trans-signalling mice, which were genetically engineered to execute IL-6 trans-signalling, with a strategy that mimics ADAM10/17 hyperactivation generated by Cre-mediated deletion of the genetic information coding for the transmembrane and intracellular domain of the IL-6R. Due to this lack, IL-6R is directly secreted as soluble IL-6R. Consequently, these mice selectively execute trans-signalling, whereas classic signalling is abrogated (21). Here, we analyzed IL-6R deficient mice in diet-induced obesity and physical exercise. Albeit also IL-6 deficient mice have shown contradictory results, IL-6R deficient mice phenotypically strongly deviate from IL-6 deficient mice in wound healing (22) but not in liver regeneration (21). We were, therefore, interested if IL-6R deficient mice phenocopy IL-6 deficient mice in diet-induced obesity. Furthermore, we hypothesize that IL-6 trans-signalling is mainly involved in glucose metabolism regulation in diet-induced obesity and may be affected by physical activity, while classic signalling is linked to a homeostatic regulation.
We first generated transgenic soluble IL-6R+/+ (sIL-6R+/+) mice, as previously described (21). Shortly, in these mice, the endogenous hyper-activation of ADAM10/17 is mimicked by Cre-mediated deletion of the genetic information coding for the transmembrane and intracellular domain of the IL-6R, reflecting a situation in which only trans-signalling is active, whereas classic signalling is abrogated. According to this mouse model, membrane-bound IL-6R is entirely converted into soluble IL-6R (sIL-6R) allowing only endogenous IL-6 trans-signalling. To endorse our model, we initially tested the plasma level of sIL-6R, using wild-type (WT) littermate controls and a mouse model with complete deficiency of IL-6R (IL-6R-KO), where both IL-6 trans and classic signalling are abrogated ( Figure 1A ) . As expected, the circulating level of sIL-6R was increased in sIL-6R+/+ mice (approximately 536 ng/ml) and completely absent in IL-6R deficient mice in comparison to the strain control mice with 12.8 ng/ml for sIL-6Rf/f (wild-type 1 for sIL-6R+/+) and 11.2 ng/ml for IL-6Rf/f (wild-type 2 for IL-6R-/-)) ( Figure 1B ). We monitored body composition in resting condition and under standard diet (SD) in IL-6 trans-signalling mice, IL-6R deficient mice and the appropriate wild-type littermates. After 18 weeks, we did not observe any differences in body weight, body fat and lean mass due to different genotypes but a similar progressive tendency throughout the 12 weeks ( Figures 1C-H ). Next, we tested the effect of sIL-6R over-expression and IL-6R deficiency on glucose and insulin tolerance under standard diet by intraperitoneal glucose tolerance test after 16 h fasting at 12 weeks and at older age of 20 weeks. There was no detectable variation in glucose disposal related to different IL-6 signalling activation ( Figures 2A-H ). We also measured blood glucose levels following an insulin injection for 15, 30, and 60 minutes, with no differences observed for the different genotypes in insulin sensitivity ( Figures 2I-L ). In summary, we did not observe any influence of specific activation of IL-6 trans-signalling and deficiency of IL-6R classic and trans-signalling on body weight gain, glucose and insulin tolerance following standard diet.
In human and animal trials, indirect calorimetry is frequently used to assess the total energy expenditure together with the respiratory exchange ratio (RER), calculated as ratio between O2 and CO2 consumption, index of glucose, protein or fat oxidation as a fuel source (23). We therefore tested the basal metabolic capacity in sIL-6R+/+ and IL-6R deficient mice. As expected, we noticed a slight increase of RER during the dark phase, index of a metabolic shift towards glucose oxidation ( Figures 3A-D ). Despite a decreased respiratory quotient in sIL-6R+/+ in comparison to WT mice in both light and dark phases, no correlation with different genotypes was statistically significant. Whole carbohydrate oxidation rates were partially elevated while fatty acids oxidation rates were moderately reduced in appropriate WT controls compared to IL-6R deficient and sIL-6R+/+ mice, but the differences were not statistically significant ( Figures 3B, C, E, F ). Whole carbohydrates and fatty acids oxidation rate were calculated as previously described (24). From these data we conclude that nor IL-6 classic or trans-signalling is influencing the basal metabolism under resting conditions.
It has been shown that IL-6 has an important role in obesity and insulin resistance but also is cardinal in driving the beneficial effects of physical exercise in glucose and insulin sensitivity as well as body composition (1). Therefore, sIL-6R trans-signalling mice, IL-6R deficient mice and appropriate wild-type mice were fed with high fat diet (HFD) starting from week 4 until week 18 including a treadmill protocol after 9 weeks of high fat diet for the total duration of 6 weeks, 5 days per week. HFD feeding determined an increase in body weight and changes in body composition trajectory in comparison to SD. A similar trend was observed in body weight and body composition before starting the treadmill training. After 6 weeks of training, a significant increase of body weight was observed in appropriate WT mice compared to trained WT and IL-6R deficient mice, due to beneficial effect of exercise, but no differences were attributable to different genotype ( Figure 4A ). After training, we observed a reduction in body fat mass but not in lean mass ( Figures 4B, C ). We did not observe any significant change in body weight, body fat and lean mass in sIL-6R+/+ and control mice after training ( Figures 4D-F ). Taken together, no differences in energy metabolism due to IL-6R genotype were underlined following high-fat diet and physical exercise.
IL-6 signalling has been reported to be involved not only in regulation of weight loss but also in improved glucose and insulin sensitivity induced by physical activity (14). Considering that, fasting blood glucose was measured at the beginning of the HFD at 4 weeks and no variations between genotypes were observed ( Figures 5A, B ). We repeated the same experiments at the beginning of the training at 12 weeks and similarly glucose level was not impaired ( Figures 5C, D ). HFD led to an increased blood glucose level after intraperitoneal glucose tolerance test measured at 11-week-old before training compared to SD-fed mice but no significant differences were observed between the genotypes ( Figures 5E-H ). To test whether exercise training has any impact, equally, glucose in addition to insulin tolerance was determined in high-fat diet-fed and trained sIL-6R+/+ and IL-6 deficient mice and littermate controls. Physical exercise significantly improved glucose tolerance in IL-6R littermate WT mice but only slightly in IL-6 deficient mice ( Figures 5I, J ), while no differences were observed between sIL-6R+/+ and littermate control mice ( Figures 5K, L ). Likewise, insulin tolerance after treadmill training in IL-6R deficient and sIL-6R+/+ mice was comparable to the sedentary group ( Figures 5M–P ). In brief, lack of membrane-bound and soluble IL-6R did not regulate fast blood glucose level following 8 weeks of high-fat diet and did not influence glucose and insulin tolerance tests during high-fat diet and after physical exercise.
To better understand the consequences of high-fat diet as well as the impact of physical exercise on the whole-body metabolism in sIL-6R+/+ and IL-6 deficient mice, we assessed substrate utilization and whole-body carbohydrate and fatty acid oxidation rate two weeks before the start of the treadmill training and after 4 weeks of exercise. We found that substrate utilization was not influenced by physical exercise and was not different between sIL-6R+/+ and IL-6R deficient mice and respective WT mice ( Figures 6A, D, G, H, K, L ). Based on our results shown in Figures 6B, C, E, F, I, J, M, N , IL-6R signalling did not regulate the whole-body carbohydrate and fatty acid oxidation and they are not significantly influenced by exercise. In conclusion, the respiratory quotient and the carbohydrate and fatty acid basal metabolism are not changed in high-fat diet fed IL-6R deficient mice and following treadmill training.
Using IL-6R deficient and sIL-6R+/+ trans-signalling mice, we analysed the specific role of IL-6 classic and trans-signalling in diet-induced obesity and physical exercise. In contrast to previous studies, mainly centred around IL-6 KO mouse models, our data suggested that both classic and trans-signalling have no considerable impact on body weight increase and distorted glucose metabolism following standard CHO diet and/or high-fat diet. Interestingly, sIL-6R+/+ mice showed modest increase in body weight in comparison to controls under CHO diet, majorly attributable to enhanced body lean mass; this phenomenon could be partially due to the observed gain of fatty acid utilization during light and dark periods that could explain the lack of increase in adipose tissue. Moreover, we did not observe any impaired blood glucose level after 9 weeks of HFD in IL-6R deficient mice and respective wild-type control. Lastly, deficiency of IL-6R did not critically regulate the beneficial effects of physical exercise on body weight, body composition, glucose and insulin tolerance, as well as basal metabolism. For the first time in the context of obesity, we characterized the sIL-6R+/+ mouse model (21), an approach to simulate ADAM10/17 hyperactivation for the specific target protein IL-6R by a Cre-mediated deletion of the genetic sequence coding for transmembrane and intracellular domains that consequently lead to specific trans-signalling execution. This is a unique system to study solely the IL-6 trans-signalling, without the classic signalling activation, during patho-physiological conditions, including, in our case, obesity associated with insulin signalling and glucose metabolism distortion, as well as physical exercise. Based on our data, following high-fat diet, circulating sIL-6R is increased up to 48-fold in sIL-6R+/+ mice compared to appropriate littermate controls, suggesting that our model specifically executes trans-signalling and abrogates classic signalling. To support our statement, we have shown that mRNA levels of IL6-R in sIL-6Rfl/fl and IL-6Rfl/fl were comparable to wild-type mice while a significant increase was detected in sIL-6R+/+, and immunochemistry data additionally highlighted IL-6R being detectable in wild-type but not in IL-6R deficient and sIL-6R+/+ mice, demonstrating how membrane-bound IL-6R is being converted to sIL-6R and rapidly secreted (21). Collectively, this explains the notable increase of sIL-6R levels in our mouse model. The present study was designed to clarify whether IL-6R deficient mice phenocopy the IL-6 deficient mice in diet-induced obesity and physical exercise exposure. The majority of studies linked IL-6 deficiency to development of obesity, glucose intolerance and insulin resistance, such as Wallenius et al., who indicated IL-6 KO mice to developed mature-onset obesity and insulin resistance (2), although a few years later Di Gregorio et al. reported no differences in obesity in 8 months old mice and after 3 months of high-fat diet (25). Remarkably, our data did not show a similar phenotype between IL-6 KO and IL-6 receptor KO mice, since we did not observe any increase in body weight, changes in body composition or alterations of glucose and insulin sensibility or variations in basal metabolism, e.g. indirect calorimetry measurements. A similar scenario has been observed by McFarland-Mancini et al., where mice lacking IL-6R showed different phenotype compared to IL-6-deficient mice in delayed wound healing process, despite some similarities in inflammatory deficits (22). Of note, in this study, the combined deficiency of IL-6 and IL-6R had the similar phenotype than IL-6R. Although we did not provide comparative data for IL-6 KO, due to the high number of studies already reported in literature, we speculate that IL-6R KO plays a minor role in development of the phenotypes compared to IL-6 KO. On this basis, a plausible explanation of our adverse results could be a previously unidentified function of IL-6R that does not involve IL-6 but other cytokines or receptors. This would highlight the feasibility of the hypothesis previously formulated by McFarland-Mancini et al. that IL-6 might execute its function binding to a different receptor. Furthermore, one cannot exclude a compensatory mechanism that involves other cytokines, belonging to the IL-6 family. For instance, Schuster et at. revealed that human CNTF as well can bind and activate signalling via IL-6R, although with an affinity 50-fold inferior to IL-6 (26). Another important factor to be considered is the genetic background differences that could affect the results. There are many examples of how genetic background could cause opposite effects on metabolism, for instances different murine strains have opposing effects on muscle and liver insulin sensitivity (27). It cannot be excluded that there are additional genetic variations into the mouse lines, diverse breeding strategies, divergent use of control animals (littermates or general WT), different age, as well as different methodology, environmental or dietary factors that could play a key role in causing variances between the studies. Notably, we observed surprising differences between the two control strains IL-6Rf/f and sIL-6Rf/f with respect to body weight and body composition, as well as blood glucose levels and RQ, equally in standard condition and after high-fat diet and physical exercise, independent of IL-6R genotype, with the same C57/BL6 background and identical experimental techniques. This underlines the importance of comparing the results with littermate controls rather than universal WT mice. Together with genotype and environmental factors, an additional variable that could influence the different pathophysiology of obesity in different mice lines is the gut microbiota. It has been shown that diet-induced obese mice have different amounts of the two dominant bacterial divisions, the Bacteroidetes and the Firmicutes, and this affects the metabolic profile, leading to increased capacity for dietary energy harvest and higher body fat contents (28). In summary, nor membrane-bound IL-6R and/or sIL-6R seems to be involved in regulating glucose and insulin metabolism in diet-induced obesity, although previous reports defined IL-6 as the major trigger of inflammation in adipose tissue in obese conditions and consequent glucose and insulin intolerance in peripheral tissues. Our results add new crucial information to what is the already known as the complex scenario related to the role of IL-6 signalling in the above mentioned mechanisms. This paves the way to address new questions that need further investigations regarding a plausible novel mechanism of IL-6R in adipocytes and skeletal muscles.
All experiments were approved by the Ethics Committee of the State Office for Nature, Environment and Consumer Protection NRW (LANUV, North Rhine-Westphalia, Germany – 84-02.04.2020.A278) and conducted at the animal facility of the German Diabetes Center. Unless otherwise specified, three to six mice per cage were housed at 22°C with a 12-hour light–dark cycle and free access to food and water. Following weaning, male animals were fed a regular chow diet (Ssniff, Soest, Germany, 11% Fat) or fed with high fat diet (HFD, Research diet, rodent, 60 kcal% Fat) from the age of 4 weeks and used for experiments from the age of 4 to 21 weeks. Floxed sIL-6R and global sIL-6R+/+ mice were generated as previously described (21). IL-6R-/- mice (22) were obtained from the Heinrich-Heine University of Düsseldorf’s animal facility ZETT – Zentrale Einrichtung für Tierforschung und wiss. Tierschutzaufgaben.
Body weight was measured every 2-3 weeks with an electronic scale (Sartorius), and body composition (body fat and lean mass) was determined by a nuclear magnetic resonance spectrometer (Bruker-Minispec NMR-Analyzer mq10; Bruker Optics).
Mice fasted for 6 h before the experiments. Fast blood glucose level was measured from the tail tip. This experiment has been performed at the age of 4 weeks and 12 weeks.
Mice were fasted for 16 h before the experiment and subsequently were injected intraperitoneally with sterile glucose (2 g/kg body weight, 20% solution, 10 µL/g). Basal blood glucose was determined at the tail tip at 0, 15, 30, 60, 120 and 240 minutes after injection. This experiment has been performed at the age of 12 and 20 weeks, under CHO-diet, and at the age of 11 weeks (pre-exercise) and 17 weeks (post-exercise), under high fat diet.
Subsequently to blood glucose level measurement from the tail tip, mice were injected intraperitoneally with 10 µL/g insulin (Actrapid, Novo Nordisk, 100 U/ml). Basal blood glucose was determined from the tail tip at 0, 15, 30 and 60 min after injection.
After a 24-hour adaption period, the respiratory quotient (RQ) of the animals was assessed using indirect calorimetry (Hartmann & Braun). The flow rate was 0.5 L/min, and the rates of oxygen consumption (VO2) and carbon dioxide production (VCO2) were measured at 22°C for 23 h. Water and food were freely available to the animals. The RQ is the quotient of VCO2/VO2. The following formulae were used to compute whole-body carbohydrate and fat oxidation rates (g/min): ; (24).
To quantify mouse serum sIL-6R, the enzyme-linked immunosorbent assay (Mouse IL-6Ra DuoSet, cat. #DY1830, R&D Systems, Minneapolis, MN, USA) was used. Microtitre plates (Nunc maxi sorb, Sigma Aldrich, Munich, Germany) were incubated overnight with goat anti-mouse IL-6R capture antibody diluted in PBS (R&D Systems, 1.6 g/ml working concentration). After overnight incubation, the plates have been washed three times with 300 µL washing buffer (R&D Systems, cat. #WA12) prior blocking with 300 µL of PBS with 1% BSA for 1h at RT. Subsequently, diluted serum samples from sIL-6Rfl/fl, sIL-6R+/+, IL-6R-/- mice and wt littermates were added (100 µL/well) and incubate for 2h at RT. Plates were washed three times and biotinylated goat anti-mouse IL-6R mAbs (R&D Systems) were used to identify bound sIL-6R.
Mice aged 12 to 18 weeks ran on a treadmill 5 days a week for 6 weeks. The treadmill program was designed to gradually increase the intensity and duration of exercise, beginning with a warm-up and 10 m/min for 20 minutes totally (0° inclination) and progressing to 15-22 m/min for a total of 60 min of training with a 10° inclination.
Data are described as means ± SEMs. Significant differences were determined using one-way ANOVA, as indicated in the figure legends. P value <0.05 were considered statistically significant.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
All experiments were approved by the Ethics Committee of the State Office for Nature, Environment and Consumer Protection NRW (LANUV, North Rhine-Westphalia, Germany – 84-02.04.2020.A278) and conducted at the animal facility of the German Diabetes Center.
ARM. carried out the experiments and wrote the manuscript with support from JS. JS conceived the original idea and supervised the project. PR helped with mouse experiments and contributed to the project discussion. AC and HA-H helped in supervising the project. KB contributed to preparation of animal documentation. All authors provided constructive criticism and contributed to the development of the study. All authors contributed to the article and approved the submitted version.
Funded by the Deutsche Forschungsgemeinschaft, Graduiertenkolleg VIVID.
The authors thank RTG 2576 “Vivid- In vivo investigations towards the early development of type 2 diabetes” for funding this project. We thank Lena Espelage, Anna Scheel, Carina Heitmann and Birgit Knobloch for technical assistance.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647095 | Li Chen,Xintong Kang,Xiujuan Meng,Liang Huang,Yiting Du,Yilan Zeng,Chunfeng Liao | MALAT1-mediated EZH2 Recruitment to the GFER Promoter Region Curbs Normal Hepatocyte Proliferation in Acute Liver Injury | 15-04-2022 | MALAT1,EZH2,GFER,H3K27me3,Methylation,Acute liver injury | Background and Aims The goal of this study was to investigate the mechanism by which the long noncoding RNA MALAT1 inhibited hepatocyte proliferation in acute liver injury (ALI). Methods Lipopolysaccharide (LPS) was used to induce an ALI cellular model in HL7702 cells, in which lentivirus vectors containing MALAT1/EZH2/GFER overexpression or knockdown were introduced. A series of experiments were performed to determine their roles in liver injury, oxidative stress injury, and cell biological processes. The interaction of MALAT1 with EZH2 and enrichment of EZH2 and H3K27me3 in the GFER promoter region were identified. Rats were treated with MALAT1 knockdown or GFER overexpression before LPS induction to verify the results derived from the in vitro assay. Results MALAT1 levels were elevated and GFER levels were reduced in ALI patients and the LPS-induced cell model. MALAT1 knockdown or GFER overexpression suppressed cell apoptosis and oxidative stress injury induced cell proliferation, and reduced ALI. Functionally, MALAT1 interacted directly with EZH2 and increased the enrichment of EZH2 and H3K27me3 in the GFER promoter region to reduce GFER expression. Moreover, MALAT1/EZH2/GFER was activated the AMPK/mTOR signaling pathway. Conclusion Our study highlighted the inhibitory role of reduced MALAT1 in ALI through the modulation of EZH2-mediated GFER. | MALAT1-mediated EZH2 Recruitment to the GFER Promoter Region Curbs Normal Hepatocyte Proliferation in Acute Liver Injury
The goal of this study was to investigate the mechanism by which the long noncoding RNA MALAT1 inhibited hepatocyte proliferation in acute liver injury (ALI).
Lipopolysaccharide (LPS) was used to induce an ALI cellular model in HL7702 cells, in which lentivirus vectors containing MALAT1/EZH2/GFER overexpression or knockdown were introduced. A series of experiments were performed to determine their roles in liver injury, oxidative stress injury, and cell biological processes. The interaction of MALAT1 with EZH2 and enrichment of EZH2 and H3K27me3 in the GFER promoter region were identified. Rats were treated with MALAT1 knockdown or GFER overexpression before LPS induction to verify the results derived from the in vitro assay.
MALAT1 levels were elevated and GFER levels were reduced in ALI patients and the LPS-induced cell model. MALAT1 knockdown or GFER overexpression suppressed cell apoptosis and oxidative stress injury induced cell proliferation, and reduced ALI. Functionally, MALAT1 interacted directly with EZH2 and increased the enrichment of EZH2 and H3K27me3 in the GFER promoter region to reduce GFER expression. Moreover, MALAT1/EZH2/GFER was activated the AMPK/mTOR signaling pathway.
Our study highlighted the inhibitory role of reduced MALAT1 in ALI through the modulation of EZH2-mediated GFER.
Acute liver injury (ALI) is a pernicious clinical condition marked by rapid hepatocyte dysfunction and defects in patients with or without a history of liver disease.1 Hepatitis viruses, drugs, immunologic injury and other hepatotoxic factors cause significant hepatocyte death, ultimately inducing ALI or even acute liver failure (ALF).2 Increases in aspartate aminotransferase (AST) and alanine aminotransferase (ALT) are used to identify the likelihood of ALI.3,4 Currently, immunosuppressors, antiviral agents, bioartificial livers, and liver transplantation are available treatment options for ALI.5 For patients with ALF and acute-on-chronic liver failure, the only definitive option is liver transplantation when there is a poor prognosis.6 Of note, the prognosis is often made worse by ineffective treatment, high cost, risk of organ rejection, limited liver donors, and severe treatment-related adverse effects.5,7 Under the circumstances, it is imperative to develop novel treatment strategies to prevent ALI. Long noncoding (lnc)RNAs are transcripts >200 nucleotides that are dysregulated in liver disease and are considered biological markers for the diagnosis, prognosis, and treatment.8,9 Abnormal expression of MALAT1 has been identified in rodent models and patients with acute kidney injury.10 Interestingly, downregulation of MALAT1 blocks hypoxia/reoxygenation-induced hepatocyte apoptosis and limits the release of lactic dehydrogenase (LDH).11 Knockdown of MALAT1 improves the outcome of lipopolysaccharide (LPS)-induced acute lung injury and suppresses apoptosis of human pulmonary microvascular endothelial cells.12 Moreover, MALAT1 recruits the histone methyltransferase EZH2 to the microRNA (miR)-22 promoter region, thus inhibiting the expression of miR-22.13 Furthermore, MALAT1 recruits EZH2 to the promoter region of ABI3BP to downregulate its expression and modulate H3K27 methylation in gallbladder cancer cells.14 Notably, regulation of methylation plays an essential role in the deterioration and management of ALI.15,16 In addition, GFER encodes augmenter of liver regeneration (ALR) a protein that specifically supports liver regeneration.17 Transient knockdown of GFER has been found to promote cell death and reduce cell proliferation in liver tissue.18 We hypothesized that GFER is a downstream gene regulated by MALAT1 in ALI. The study aim was to elucidate the role and potential mechanism of MALAT1 in LPS-induced ALI. To that end, animal and cellular models of ALI were established using LPS, and the severity of ALI, apoptosis and proliferation were assessed.
This study included 26 patients with acute drug-induced liver injury (ADILI) who were hospitalized in the liver disease department of The Third Xiangya Hospital of Central South University and 19 healthy people from the physical examination clinic of The Third Xiangya Hospital of Central South University. The diagnostic criteria of ADILI were based on the guidelines for the diagnosis and treatment of drug-induced liver injury of the 17th National Conference on Viral Hepatitis and Hepatology of the Chinese Medical Association in 2015 and a Roussel Uclaf Causality Assessment Method (RUCAM) scale score of >8 points.19 Patients were excluded if they had viral hepatitis B or C or other types of viral hepatitis, autoimmune liver disease, alcoholic liver disease, nonalcoholic fatty liver disease, cholestatic disease, or inherited metabolic liver disease. Peripheral blood was selected for clinical research in this study because liver biopsies were not performed. Approximately 4 mL of peripheral blood was collected in the morning after overnight fasting, placed in a serum tube, centrifuged at 4,000 rpm at 4°C for 10 m, and stored at −80°C until. use. This study complied with the Declaration of Helsinki and was approved by the ethics committee of The Third Xiangya Hospital of Central South University. All patients signed an informed consent form (No. 22014). Patient information is shown in Table 1.
HL7702 human hepatocyte cells were provided by Cell Bank of Chinese Academy of Science and maintained in Dulbecco’s modified Eagle’s medium (Gibco, Grand Island, NY, USA) containing 10% fetal bovine serum plus 1% streptomycin-penicillin and cultured in a 37°C incubator with 5% CO2. Hepatocytes were induced by 1 µg/mL LPS (Sigma-Aldrich, St Louis, MO, USA) for 16 h in a 5% CO2 atmosphere at 37°C to induce ALI. In some cases, hepatocytes were cultured with an adenosine monophosphate-activated protein kinase (AMPK) inhibitor (10 µM, Compound C; Sigma-Aldrich) for 1 h before LPS induction. HL7702 cells were seeded in culture plates and transfected with MALAT1 overexpression vector (oe-MALAT1), MALAT1 knockdown vector (sh-MALAT1, 20 µL, viral titer 5×108 TU/mL), EZH2 overexpression vector (oe-EZH2, 20 µL, viral titer 5×108 TU/mL), EZH2 knockdown vector (sh-EZH2, 20 µL, viral titer of 5×108 TU/mL), GFER overexpression vector (oe-GFER, 20 µL, viral titer of 3×108 TU/mL), GFER knockdown short hairpin vector (sh-GFER, 20 µL, viral titer of 5×108 TU/mL) or the corresponding negative controls (oe-NC and sh-NC) (20 µL, 3×108 TU/mL). Lentivirus vectors used for gene overexpression (LV5-GFP) or knockdown (pSIH1-H1-copGFP) were provided by GenePharma (Shanghai, China). Each experiment was conducted in triplicate. LPS induction was performed in HL7702 cells 24 h after transfection.
The animal procedures in this study were approved by the Ethics Committee of The Third Xiangya Hospital of Central South University (No. 22014) and followed the guidelines of the National Institutes of Health. Forty-two Sprague-Dawley rats 7–8 weeks of age, and 200–250 g (Shanghai Laboratory, Animal Research Center, Chinese Academy of Science) were fed under pathogen-free conditions and kept in a 12 h light/dark cycle and 60–65% humidity. The rats were allowed free access to food and water. The rat ALI model was induced by intraperitoneal injection of 10 mg/kg LPS (Sigma-Aldrich), with normal saline treatment as the control. The model was maintained for 6 h before sh-MALAT1 (5×108 TU/mL, 300 µL/rat) or oe-GFER (3×108 TU/mL, 300 µL/rat) was introduced into rats through intravenous injection in tails for MALAT1 knockdown or GFER overexpression, with sh-NC (3×108 TU/mL, 300 µL/rat) or oe-NC (3×108 TU/mL, 300 µL/rat) as the negative control. The 42 rats were randomly divided into seven groups of six rats each, including normal controls, and saline (administration of an equal volume of normal saline), LPS (intraperitoneal injection of 10 mg/kg LPS for 6 h modelling), LPS + sh-NC group (intravenous injection of sh-NC in the tail vein 42 h before LPS induction, followed by intraperitoneal injection of 10 mg/kg LPS for 6 h modelling), LPS + sh-MALAT1 (injection of sh-MALAT1 in the tail vein 42 h before LPS induction, followed by intraperitoneal injection of 10 mg/kg LPS for a 6-h modelling), LPS + oe-NC group (injection of oe-NC in the tail vein 42 h before LPS induction, followed by intraperitoneal injection of 10 mg/kg LPS for a 6-h modelling), and LPS + oe-GFER (injection of oe-GFER in the tail vein 42 h before LPS induction, followed by intraperitoneal injection of 10 mg/kg LPS for a 6-h modelling). All animals were sacrificed to collect serum and liver tissue 48 h after modelling for 6 h and injection of lentivirus vectors.
RNA extraction of cells or tissues was carried out using TRIzol (Invitrogen, Carlsbad, CA, USA) followed by detection of RNA concentration and purity. Qualified RNA samples were adjusted to an appropriate concentration and then reverse-transcribed using random primers and a reverse transcription kit (TaKaRa, Tokyo, Japan) following the manufacturer’s instructions. Gene expression was quantified by fluorescence qRT-PCR (LightCycler 480; Roche, Indianapolis, IN, USA), and was carried out following the manufacturer’s instructions (SYBR Green Mix; Roche Diagnostics). In brief, cDNA templates were predenatured at 95°C for 5 m, denatured at 95°C for 10 s, annealed at 60°C for 10 s and finally extended at 72°C for 20 s, followed by 40 cycles of cycling. Each qPCR assay was performed with three replicates. The relative expression of GFER and MALAT1 was determined by the 2−ΔΔCt method with GAPDH as the reference gene. The primer sequences for GAPDH, MALAT1, and GFER are shown in Table 2.
Cells or tissues were lysed in RIPA lysis buffer and centrifuged for protein extraction. Protein concentration was detected with a bicinchoninic acid assay kit (Beyotime Biotechnology, Shanghai, China) to ensure equal loading volume of protein. The corresponding volume of protein was homogenized with loading buffer (Beyotime) and then denatured for 3 m in a boiling water bath. Proteins were separated on 10% sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gels following the kit (Beyotime) manufacturer’s instructions. Briefly, the protein was electrophoresed at 80 V, and then the voltage was switched to 120 V for 1–2 h. After protein separation, membrane transfer was performed in an ice bath at 300 mA for 60 m. Then, the membranes were rinsed in a washing solution for 1–2 m, followed by blocking at room temperature for 60 m or at 4°C overnight. Incubation with primary antibodies against rabbit anti-human GAPDH (1:1,000; Cell Signaling, Boston, MA, USA), GFER (1:500; Santa Cruz Biotechnology, Dallas, TX, USA), p-mTOR (Ser2448 1:1,000; Cell Signaling Technology), mTOR (1:1,000; Cell Signaling Technology), p-AMPK (Thr172 1:1,000; Cell Signaling Technology), AMPK (1:1,000; Cell Signaling Technology), H3K27me3 (1:1,000; Abcam, Cambridge, UK), or EZH2 (1:500; Abcam) performed at room temperature on a shaking table for 1 h. After incubation, the membranes were washed three times for 10 m each and then incubated with horseradish peroxidase-labeled goat anti-rabbit IgG (1:5,000; Beijing ComWin Biotech Co., Ltd., Beijing, China) for 1 h at room temperature. Before color development, the membranes were washed three times for 10 m each. A chemiluminescence imaging system (Bio-Rad, Hercules, CA, USA) was used to visualize the membranes.
The culture supernatant of HL7702 cells or rat serum was collected for liver function testing. AST, ALT, and LDH were assayed with kits following the manufacturer’s (Sigma-Aldrich, Merck KGaA, Darmstadt, Germany) instructions.
Twenty-four hours after transfection, 100 µL of the cell suspension was seeded into a 96-well plate, with three replicates for each sample. Cells were cultured in an incubator for 24, 48, or 72 h, 10 µL of CCK-8 reagent (Dojindo, Tokyo, Japan) for 3 h, and absorbance was determined at 450 nm.
Cells (1×105 cells/well) were transferred to 96-well plates and cultured for 2 h with 100 µL EdU stain (5 µM; Sigma-Aldrich, Merck KGaA). The cells were then immobilized in 50 µL fixation buffer for 30 m. After removing the buffer, the cells were incubated with 50 µL glycine (2 mg/mL) for 5 m and washed with 100 µL phosphate buffered saline (PBS). After culture with 100 µL of permeabilization buffer and washing in 100 µL PBS, the cells were incubated with 100 µL of 1× Apollo solution in the dark for 30 m. Finally, the cells were cultured with 100 µL diamidino-phenylindole at room temperature for 5 m away from light and washed in 100 µL PBS, followed by observation by fluorescence microscopy (Olympus, Tokyo, Japan).
Collected cells were fixed in 4% paraformaldehyde for 30 m and then in 70% cold alcohol for 15 m. The cells were then incubated with PBS containing 0.3% Triton X-100 at room temperature for 5 m and incubated with TUNEL solution (Beyotime) for 60 m at 37°C. After washing in PBS three times, the cells were blocked with an antifade reagent and observed by fluorescence microscopy. The cell nuclei were stained with diamidino-phenylindole and apoptosis (%)=(TUNEL-positive cells/total cells)×100. Rats were sacrificed to collect liver tissues, which were fixed in 10% neutral buffer formalin (Beijing Solarbio Science & Technology Co., Ltd., Beijing, China) for 24 h, dehydrated, and embedded in paraffin. After being sliced into 4 µm serial sections, the tissue was dewaxed with xylene and dehydrated in an alcohol gradient. A TUNEL detection kit (ZK-8005; ZSGB-Bio, Beijing, China) was used to determine the apoptosis rate in rat liver tissues. Five random fields in each section were evaluated by light microscopy (Olympus Optical Co. Ltd., Tokyo, Japan). Apoptotic cells were brown or brownish in color and with apoptotic cell morphology. Apoptosis was reported as the apoptosis index, and apoptosis (%)=(TUNEL-positive cells/total cells)×100.
Assay kits were used to determine MDA, SOD (Abcam), and GSH (Sigma-Aldrich) levels in cultured cells and liver tissues following manufacturer’s instructions.
Magna RIP RNA-binding protein immunoprecipitation kits (Millipore Corp, Billerica, MA, USA) was used for the RIP assay. Briefly, HL7702 cells were lysed in 100 µL lysis buffer containing protease and RNase inhibitors, and then the protein lysate was incubated with rabbit anti-human EZH2 antibody (ab186006 1:500; Abcam) at 4°C for 30 m or anti-IgG antibody (ab109489, 1:100; Abcam) as the control. Subsequently, 10–50 µL of protein A/G beads were added and incubated with the cells at 4°C overnight. After incubation, the protein A/G-bead-antibody complexes were washed 3–4 times in 1 mL lysis buffer, and RNA was extracted and purified using the RNA extraction method. qRT-PCR was carried out with a MALAT1-specific primer to identify the interaction between EZH2 and MALAT1.
The ChIP assay was performed with SimpleChIP Plus sonication chromatin IP kits following the manufacturer’s (Cell Signaling Technology) instructions. HL7702 cells were fixed in 1% formalin to crosslink DNA and proteins. Then, the cells were lysed in lysis buffer and nuclear lysis buffer and ultrasonicated to generate 200–300 bp chromatin segments. The cell lysate was immunoprecipitated with protein A-beads conjugated with the corresponding antibodies, including anti-EZH2 antibody (ab228697; Abcam) and anti-histone 3 antibody (trimethyl-K27, ab6002; Abcam). Anti-IgG antibody (ab171870; Abcam) was added as a negative control. Protein-DNA crosslinking was reversed the RNA was purified, and enrichment of the DNA segment was detected by qRT-PCR.
Rat liver tissues were collected, fixed in 10% neutral buffered formalin (Beijing Solarbio Science & Technology Co., Ltd.) for 24 h, dehydrated, embedded in paraffin, cut into 4 µm serial sections, and stained with H&E (Beijing Solarbio Science & Technology Co., Ltd.). Tissue histology was evaluated by optical microscopy (Olympus Optical, Tokyo, Japan).
Collected liver tissues were fixed in 4% paraformaldehyde for 48 h, embedded in paraffin, and sectioned at 4 µm. The sections were baked for 20 m, dewaxed in xylene, and washed once in distilled water. After washing three times in PBS, the sections were placed in 3% H2O2 for 10 m and subjected to antigen retrieval. After washing with PBS three times, the sections were blocked in goat serum at room temperature for 20 m. Excess serum was discarded, and the primary antibody against Ki-67 (ab16667, 1:200; Abcam) was added to the sections for incubation (4°C, overnight). Afterwards, the sections were incubated with secondary antibody at room temperature for 1 h and washed three times in PBS. Color development was sustained for 1–3 m using diaminobenzidine solution, and H&E was then used for 3 m for nuclear staining. The sections were then dehydrated, permeabilized, and cover slipped for observation. The percentage of positive cells in five randomly selected fields was reported (positive cells (%)=(positive cells/total cells)×100.
GraphPad 7.0 software was used for the statistical analysis, and data were reported as means ± SD. The significance of between-group differences was determined by t-tests. Multiple comparisons was carried out with one-way analysis of variance followed by Tukey’s multiple comparisons test. P-values <0.05 were considered statistically significant.
To explore the expression of MALAT1 and GFER in patients with ALI, we included 26 patients with ADILI and 19 healthy individuals. There were no significant differences in sex or age between the two groups. Compared with the normal group, the serum levels of ALT, AST, γ-GT, alkaline phosphatase (ALP), and TBil, and the RUCAM scores of patients in the ADILI group increased significantly (Table 2; p<0.05). The results of qRT-PCR and western blotting showed that, compared with the normal group, the expression of MALAT1 in the serum of patients in the ADILI group increased, and the expression of GFER decreased (Fig. 1A, B; p<0.05). The results indicated that MALAT1 was highly expressed but GFER was weakly expressed in the serum of patients with ALI.
Human hepatocytes HL7702 were treated with LPS to induce an ALI cellular model, and ALT, AST, and LDH levels were assayed in the culture supernatant. LPS induction increased ALT, AST, and LDH levels (Fig. 2A, C; p<0.05). The CCK-8 assay found that proliferation was decreased in the LPS group compared with the control group (Fig. 2D; p<0.05), which was confirmed by EdU staining (Fig. 2E; p<0.05). TUNEL staining found that the cell apoptosis rate was increased in the LPS group compared with the control group (Fig. 2F, G; p<0.05). Assays of MDA, SOD, and GSH in cells found that after LPS treatment, MDA increased significantly and SOD and GSH decreased significantly (Fig. 2H, J; p<0.01). Moreover, qRT-PCR and western blot assays found that LPS increased MALAT1 expression (Fig. 2K; p<0.001) and decreased GFER expression (Fig. 2L, M; p<0.05) in HL7702 cells compared with the control group. Overall, LPS induction enhanced MALAT1 expression and reduced GFER expression in HL7702 cells.
HL7702 cells were transfected with sh-MALAT1, oe-MALAT1, sh-GFER, or oe-GFER for 24 h, followed by treatment with LPS for 16 h. The transfection efficiency was validated by qRT-PCR and western blot analysis (Fig. 3A–D; p<0.05). In addition, MALAT1 overexpression inhibited GFER expression and MALAT1 knockdown enhanced GFER expression (Fig. 3B–D; p<0.05). However, overexpression or knockdown of GFER did not influence MALAT1 expression (Fig. 3A). Introduction of MALAT1 overexpression or GFER knockdown substantially increased the levels of ALT, AST and LDH, while MALAT1 knockdown or GFER overexpression had the opposite effects (Fig. 3E–G; p<0.05). Compared with the LPS + oe-MALAT1 group, the LPS + oe-MALAT1 + oe-GFER group had decreased ALT, AST and LDH levels (Fig. 3E–G; p<0.05). Moreover, MALAT1 overexpression or GFER knockdown inhibited HL7702 cell proliferation and induced cell apoptosis and oxidative stress injury, but MALAT1 knockdown or GFER overexpression increased the proliferation rate and decreased the apoptosis rate and oxidative stress injury (Fig. 3H–M; p<0.05). In the LPS + oe-MALAT1 + oe-GFER group, HL7702 cells possessed increased proliferative ability and decreased apoptosis rate and oxidative stress injury than those in the LPS + oe-MALAT1 group (Fig. 3H–M; p<0.05). The data indicate that downregulation of MALAT1 inhibited hepatocyte apoptosis and oxidative stress injury but promoted cell proliferation by regulating GFER.
GFER significantly reduced ALT and AST in a mouse ALI model, and alleviated liver injury caused by ischemia-reperfusion.20,21 The University of California Santa Cruz (UCSU) genome browser (http://genome-asia.ucsc.edu/) predicted the presence of H3K27me3 methylation peak in the GFER promoter region (Fig. 4A). Inhibition of MALAT1 reduces liver ischemia-reperfusion injury,11 and MALAT1 changes the progression of liver fibrosis by regulating SIRT1.22 A previous study reported that MALAT1 recruited histone methyltransferase EZH2 to the pri-miR-22 promoter region to inhibit miR-22 expression.13 In this study, MALAT1 expression was negatively associated with GFER expression in the cellular ALI model, and the regulation of HL7702 cell proliferation and apoptosis by MALAT1 was involved in GFER. We hypothesized that MALAT1 recruited EZH2 to the GFER promoter region to suppress GFER expression. As expected, the RIP assay identified the interaction between MALAT1 and EZH2 (Fig. 4B; p<0.001). A ChIP assay was performed to confirm whether EZH2 regulated GFER expression via H2K27me3 methylation. EZH2 and H3K27me3 were found to be more abundant in the GFER promoter region in the oe-MALAT1 group than in the oe-NC group (p<0.01); enrichment of EZH2 and H3K27me3 in the GFER promoter region was reduced in the sh-MALAT1 group when compared with the sh-NC group (Fig. 4C; p<0.05). HL7702 cells were transfected with oe-EZH2 or sh-EZH2, and western blots of EZH2 expression demonstrated that transfection of oe-EZH2 promoted EZH2 and H3K27me3 expression and reduced GFER expression. sh-EZH2 treatment resulted in contrary findings (Fig. 4D; p<0.05). Overall, the results show that MALAT1 inhibited GFER expression by recruiting EZH2 to the GFER promoter region and promoting H3K27me3 methylation.
HL7702 cells were transfected with sh-MALAT1, oe-MALAT1, sh-EZH2, oe-EZH2, sh-GFER and oe-GFER for 24 h, followed by treatment with LPS for 16 h. The levels of phosphorylated proteins active in the AMPK/mTOR signaling pathway were assayed by western blotting. MALAT1/EZH2 overexpression or GFER knockdown significantly upregulated phosphorylated AMPK levels and downregulated phosphorylated mTOR levels in HL7702 cells;. MALAT1/EZH2 knockdown or GFER overexpression had the opposite effects (Fig. 5A–C; p<0.05). After HL7702 cells were transfected with oe-MALAT1 lentiviral vector and its control (oe-NC) for 24 h, they were treated with the AMPK inhibitor Compound C (CC) for 1 h, induced by LPS for 16 h, and then collected for subsequent assays. Western blot assays demonstrated that compared with the LPS + oe-NC group, phosphorylated AMPK level was decreased and phosphorylated mTOR expression was increased in the LPS + oe-NC + CC group. HL7702 cells in the LPS + oe-MALAT1 group had the reverse responses; phosphorylated AMPK level was reduced and that of mTOR was increased in the LPS + oe-MALAT1 + CC group compared with the LPS + oe-MALAT1 group (Fig. 5D–F; p<0.01). In the LPS + oe-NC + CC group, the levels of ALT, AST and LDH were decreased (Fig. 5G–I; p<0.05), HL7702 cell proliferation was increased (Fig. 5J, K; p<0.05), and the apoptosis rate and oxidative stress injury were inhibited (Fig. 5L–O; p<0.05) compared with those in the LPS + oe-NC group. HL7702 cells in the LPS + oe-MALAT1 group had increased ALT, AST, and LDH levels (Fig. 5G–I; p<0.05) concurrent with decreased cell proliferation (Fig. 5J, K; p<0.05) and enhanced apoptosis rate and oxidative stress injury (Fig. 5L–O; p<0.05), compared with those in the LPS + oe-NC group. In the LPS + oe-MALAT1 + CC group, the levels of ALT, AST and LDH in HL7702 cell supernatant were suppressed (Fig. 5G–I; p<0.05) along with increased proliferation (Fig. 5J, K; p<0.05) and reduced apoptosis rate and oxidative stress injury (Fig. 5L–O; p<0.05). The data indicate that MALAT1/EZH2/GFER activated the AMPK/mTOR signaling pathway.
Rats were intravenously injected with sh-MALAT1 or oe-GFER in the tail vein for 42 h, followed by treatment of LPS to induce ALI. We demonstrated that the levels of ALT, AST and LDH in rat serum were increased in the LPS group (p<0.01, vs. the saline group), while those of the LPS + sh-MALAT1 group and the LPS + oe-GFER group were lower than those in the LPS + sh-NC group and the LPS + oe-NC group (Fig. 6A–C). As shown by H&E staining, rats in the normal group and the saline group had normal histology with clear hepatic lobules and regular hepatic sinusoids. In contrast, rats in the LPS group had significant liver injury, manifested as cellular edema, hematolysis, diffuse necrosis, and inflammatory cell infiltration (Fig. 6D). However, LPS-induced injury was alleviated in the liver tissues of rats in the LPS + sh-MALAT1 and LPS + oe-GFER groups (Fig. 6D). qRT-PCR and western blotting revealed an increase in MALAT1 expression (Fig. 6E; p<0.01) and a decrease in GFER expression (Fig. 6F, G; p<0.05) in the LPS group compared with those in the saline group. In the LPS + sh-MALAT1 group, MALAT1 expression was significantly reduced (Fig. 6E; p<0.01) and GFER expression was upregulated (Fig. 6F, G, p<0.05) in liver tissues, compared with the LPS + sh-NC group. Moreover, increased GFER expression (p<0.001) and unchanged MALAT1 expression (p > 0.05) were found in the LPS + oe-GFER group compared with the LPS + oe-NC group (Fig. 6E–G). Taken together, downregulation of MALAT1 promoted GFER expression in ALI-model rats.
Rats were injected with sh-MALAT1 or oe-GFER in the tail vein for 42 h, followed by LPS induction for modelling, and then immunohistochemistry was performed. Mice in the LPS group had fewer Ki-67-positive liver cells than the saline group, but MALAT1 knockdown or GFER overexpression increased the percentage of Ki-67-positive cells in liver tissues (Fig. 7A; p<0.05). In addition, TUNEL staining indicated increased apoptosis rate in rat liver tissue from the LPS group (p<0.01, vs. the saline group) and decreased hepatocyte apoptosis in the LPS + sh-MALAT1 or the LPS + oe-GFER group (both p<0.05) (Fig. 7B). We also found that MDA was increased and SOD and GSH were decreased in liver tissue from the LPS group relative to the saline group. In the LPS + sh-MALAT1 group or the LPS + oe-GFER group, MDA level was decreased and SOD and GSH levels were increased (Fig. 7C–E; p<0.01). In addition, liver tissues in the LPS group were found to have upregulated levels of phosphorylated AMPK (Fig. 7F, G; p<0.01) and downregulated levels of phosphorylated mTOR (Fig. 7F. H; p<0.05). However, MALAT1 knockdown or GFER overexpression reduced phosphorylated AMPK and increased phosphorylated mTOR levels in liver tissue (Fig. 7F–H; p<0.05). Thus, these data proved the ameliorative effect of MALAT1 downregulation or GFER overexpression on ALI in vivo.
ALI is a severe, acute disease with a high mortality. It is characterized by acute hepatocyte necrosis, and there have not been any treatment breakthroughs in the last few decades.23,24 Recently, various lncRNAs, including DINO, NEAT1, and XIST, and their downstream genes have been investigated to explain the progression and improve the effectiveness of ALI treatment.16,25,26 LPS, which was first found in the outer membrane of gram-negative bacteria, has been widely used in vivo and in vitro to mimic the pathology of ALI.27–29 In this study, we successfully established cellular and animal models of ALI using LPS, and MALAT1 was found to inhibit cell proliferation and promote apoptosis in LPS-induced rats and hepatocytes. MALAT1 has been found to limit proliferation and induce apoptosis of human renal tubular epithelial cells in the presence of LPS,30 and MALAT1 knockdown was found to block LPS-induced acute lung injury by inhibiting apoptosis and improving cell viability.31 The apoptosis-promoting effect of MALAT1 has been previously reported in hepatocytes.31 However, Li et al.32 reported that MALAT1 overexpression was required for accelerating hepatocyte proliferation and liver regeneration. Our in vivo and in vitro experiments showed that MALAT1 was upregulated after LPS exposure and identified as an ALI-promoting gene. GFER, also known as ALR, protects against chemical- or toxin-related ALI,33,34 ischemia reperfusion-induced liver and kidney injury.35,36 Deletion of ALR accelerates steatohepatitis and hepatocellular carcinoma.37 Therefore, we examined the expression of GFER in LPS-induced hepatocytes, which showed a decrease in GFER expression. Overexpression of GFER alleviated hepatocyte apoptosis, excessive hepatocyte proliferation, and improved liver function after LPS insult. Intriguingly, MALAT1 was negatively associated with, and significantly regulated, GFER expression at the cellular and animal levels. In addition, the ALI-promoting effect of MALAT1 was offset by GFER upregulation, indicated by decreases in serum AST/ALT, hepatocyte apoptosis, and enhanced proliferation of hepatocytes. The results indicated that MALAT1 regulated GFER expression to aggravate ALI. We then focused on elucidating the mechanism of MALAT1 regulation of GFER in ALI. MALAT1 has oncogenic activity through EZH2-mediated suppression of miR-217 expression in lung carcinogenesis.38 In the presence of MALAT1, silencing of EZH2 has been shown to reduce apoptosis and enhance cardiac function.13 Increased MALAT1 recruits EZH2 to its downstream gene to promote H3K27me3 expression for specific transcription inhibition.39 EZH2 can modify the enrichment of its catalyzed H3K27me3 and contribute to the pathogenesis of liver failure by triggering the release of tumor necrosis factor (TNF) and other pro-inflammatory cytokines.40 Inhibition of EZH2 blunts H3K27me3 and restrains the activities of serum ALT and AST.41 We predicted an H3K27me3 methylation peak in the GFER peak, indicating that MALAT1 might regulate GFER expression through methylation. Our functional experiments identified an interaction between MALAT1 and EZH2, and further demonstrated that MALAT1 overexpression significantly enhanced the enrichment of EZH2 and H3K27me3 in the GFER promoter region. The study firstly demonstrated that MALAT1 suppressed GFER expression by recruiting EZH2 to the GFER promoter region, enhancing H3K27me3 methylation, and aggravating ALI. AMPK regulates energy homeostasis and metabolism, and mTOR is an enzyme downstream of AMPK.42 Activation of the AMPK/mTOR signaling pathway has been shown to be involved in liver diseases, including nonalcoholic fatty liver disease and ALI.43–45 More important, Pu et al.46 found that deletion of ALR induced increased AMPK phosphorylation and decreased mTORC1 phosphorylation, and increased both autophagy flux and apoptosis. In this study, phosphorylated AMPK was increased, and phosphorylated mTOR was reduced in ALI model rats. At the same time, MALAT1 inhibition or GFER overexpression decreased the expression of p-AMPK and promoted p-mTOR expression. Inhibiting the phosphorylation of AMPK by Compound C inhibited ALT, AST, and LDH, and hepatocyte apoptosis, and improved hepatocyte proliferation. Knockdown of MALAT1 or overexpression of GFER had a similar effect to that of Compound C in ALI. The AMPK/mTOR pathway is a typical regulator of autophagy.47 Both endogenous and exogenous interleukin-37 protect against ischemia reperfusion-induced hepatic injury by restraining excessive autophagy and apoptosis by regulating the AMPK/mTOR signaling pathway.48 Therefore, much attention should be paid in future studies to whether MALAT1/EZH2/GFER participates in ALI by activating the AMPK/mTOR pathway to affect hepatocyte autophagy. This study revealed that MALAT1 upregulation was associated with decreased proliferation, enhanced apoptosis, and aggravation of liver injury, which sheds light on the prevention and treatment of ALI. Notably, we showed that MALAT1 inhibited GFER by recruiting EZH2 to the GFER promoter region and enhancing H3K27me3 methylation, thus resulting in deterioration of ALI. Activation of the AMPK/mTOR signaling pathway was also linked to the ALI-promoting effect of MALAT1. The study thus characterized a possible regulatory mechanism of MALAT1 exacerbating ALI, which contributes to understanding ALI progression and provides a novel approach for ALI treatment. Further studies are required to explain the regulation and methylation of the molecules by MALAT1 in ALI to boost the application of their therapeutic benefits in clinical practice. |
|
PMC9647096 | Ziyue Huang,Haoming Xia,Yunfu Cui,Judy Wai Ping Yam,Yi Xu | Ferroptosis: From Basic Research to Clinical Therapeutics in Hepatocellular Carcinoma | 31-08-2022 | Hepatocellular carcinoma,Ferroptosis,Sorafenib,Cancer therapy,Molecular targets | Hepatocellular carcinoma (HCC) is one of the most common and highly heterogeneous malignancies worldwide. Despite the rapid development of multidisciplinary treatment and personalized precision medicine strategies, the overall survival of HCC patients remains poor. The limited survival benefit may be attributed to difficulty in early diagnosis, the high recurrence rate and high tumor heterogeneity. Ferroptosis, a novel mode of cell death driven by iron-dependent lipid peroxidation, has been implicated in the development and therapeutic response of various tumors, including HCC. In this review, we discuss the regulatory network of ferroptosis, describe the crosstalk between ferroptosis and HCC-related signaling pathways, and elucidate the potential role of ferroptosis in various treatment modalities for HCC, such as systemic therapy, radiotherapy, immunotherapy, interventional therapy and nanotherapy, and applications in the diagnosis and prognosis of HCC, to provide a theoretical basis for the diagnosis and treatment of HCC to effectively improve the survival of HCC patients. | Ferroptosis: From Basic Research to Clinical Therapeutics in Hepatocellular Carcinoma
Hepatocellular carcinoma (HCC) is one of the most common and highly heterogeneous malignancies worldwide. Despite the rapid development of multidisciplinary treatment and personalized precision medicine strategies, the overall survival of HCC patients remains poor. The limited survival benefit may be attributed to difficulty in early diagnosis, the high recurrence rate and high tumor heterogeneity. Ferroptosis, a novel mode of cell death driven by iron-dependent lipid peroxidation, has been implicated in the development and therapeutic response of various tumors, including HCC. In this review, we discuss the regulatory network of ferroptosis, describe the crosstalk between ferroptosis and HCC-related signaling pathways, and elucidate the potential role of ferroptosis in various treatment modalities for HCC, such as systemic therapy, radiotherapy, immunotherapy, interventional therapy and nanotherapy, and applications in the diagnosis and prognosis of HCC, to provide a theoretical basis for the diagnosis and treatment of HCC to effectively improve the survival of HCC patients.
Hepatocellular carcinoma (HCC) is the sixth most common malignancy and causes a severe burden because of its mortality, as it is the third cause of cancer-related death worldwide.1,2 Despite the gradual enhancement in treatment strategies, including surgical treatment, immunotherapy, targeted therapy, or combination therapy; the proportion of the effective population, and the availability of effective drugs and duration of efficacy, the overall survival of HCC patients is still limited.3 Studies have shown that a complex liver disease background, high tumor heterogeneity, and a high recurrence rate are major factors that limit the treatment and prognosis of HCC patients.4,5 Because of a lack of obvious clinical symptoms and early diagnostic markers at the early stage of the disease, most HCC patients cannot undergo radical surgical resection because they are in an advanced stage of the disease at the time of the visit.2,3 Sorafenib is a multikinase inhibitor that is used to treat patients with unresectable HCC.3–7 Sorafenib is the first drug used for systemic therapy of advanced HCC patients, and it effectively prolongs survival HCC, but drug resistance and adverse drug reactions limit the survival benefit. In two randomized phase III clinical trials of patients with advanced HCC, the overall response rate of sorafenib was 2–3%, the time to progression was only 5.5 months, and the median survival time was only 10.7 months6,7 Consequently, it is crucial to explore the pathogenesis and drug resistance mechanism of HCC, to further identify new therapeutic targets and develop safe and effective treatment regimens. Ferroptosis is a newly discovered special form of regulated cell death (RCD), that differs from apoptosis, necroptosis, autophagy, and pyroptosis, and is characterized by the accumulation of iron-dependent lipid peroxides.8 Accumulating evidence indicates that ferroptosis is closely related to the pathogenesis of various diseases, such as neurodegenerative diseases,9 ischemia-reperfusion injury,10 autoimmune diseases,11 liver fibrosis,12 and various cancers, including HCC.13–15 During tumorigenesis, ferroptosis has a dual role in tumor promotion and suppression, which depends on the release of damage-associated molecular patterns and the activation of immune responses triggered by ferroptotic damage within the tumor microenvironment.16 Furthermore, ferroptosis affects the efficacy of chemotherapy, radiotherapy, and immunotherapy in cancer patients.15,17,18 Therefore, although the regulatory mechanism of ferroptosis is not yet fully understood, based on its connection with various tumors, ferroptosis may become a key part of HCC treatment in the future, and some researchers also believe that ferroptosis may become a critical factor in the diagnosis and prognosis of HCC.19 Hence, this review discusses the regulatory mechanism and application of ferroptosis in HCC in detail, to provide reference for the diagnosis, treatment, and prognosis of HCC.
Ferroptosis, a form of iron-dependent RCD driven by excessive accumulation of lipid peroxidation, was first proposed by Dixon et al.8 in 2012. In contrast to apoptosis, autophagy, necrosis, pyroptosis, and other forms of programmed cell death, morphologically, the mitochondrial outer membrane of ferroptotic cells is ruptured and shrunken, and mitochondrial cristae are reduced (disappear). For cellular components, ferroptosis is often accompanied by the accumulation of iron ions, the elevation of reactive oxygen species (ROS), decreased nicotinamide adenine dinucleotide phosphate, and changes in some characteristic genes.8,20 Ferroptosis is triggered by inhibiting cell membrane translocators such as cystine/glutamate translocators (also known as system Xc-) or by activating transferrin, as well as by blocking intracellular antioxidant enzymes such as glutathione peroxidase 4 (GPX4). Additionally, iron accumulation and lipid peroxidation are two key signals that initiate membrane oxidative damage during ferroptosis (Fig. 1).8,16 Therefore, the regulation of ferroptosis mainly focuses on system Xc regulation, glutathione (GSH) metabolism and GPX4 activity regulation and iron and ROS regulation. System Xc- exchanges glutamate for cystine in a 1:1 ratio with solute carrier family 7 member 11 (SLC7A11) and transports it into the cell for GSH synthesis, and inhibition of system Xc- activity drives ferroptosis.8,21 GSH, a substrate of GPX4, protects cells from lipid peroxidative damage, and both the inhibition of GSH synthesis and the inactivation of GPX4 induce ferroptosis.22,23 Most tumors show highly invasive growth by increasing iron storage within a certain range, but excess iron concentrations lead to membrane lipid peroxidation and cell death.24,25 Moreover, the rapid proliferation of tumor cells requires the support of high mitochondrial metabolism, and mitochondria are the main source of ROS.26 Excessive ROS-induced oxidative stress damages tissues and cells. The high iron requirement, active metabolism, and increased ROS load make cancer cells more sensitive to ferroptosis. Studies have shown that ferroptosis inhibits tumor growth and promotes tumor cell chemosensitivity, so the induction of ferroptosis has promise as tumor treatment strategy.16,20 Compounds (either inducers or inhibitors) that precisely regulate ferroptosis have an important role in elucidating the mechanisms of ferroptosis-related diseases, so the identification of ferroptosis inducers or inhibitors is indispensable. Ferroptosis inducers can be divided into four categories according to their potential induction mechanisms, inhibition of System Xc- activity, direct inhibition of GPX4, depletion of GPX4 protein and CoQ10, and induction of lipid peroxidation. Ferroptosis inhibitors can be divided into two categories, those that reduce intracellular iron accumulation and those that inhibit lipid peroxidation (Table 1).8,27-40
Sorafenib is an oral multitarget, multikinase inhibitor that blocks platelet derived growth factor, vascular endothelial growth factor, Fms-like tyrosine kinase 3, c-Kit (CD117) upstream of the signal transduction pathway of tumor and tumor angiogenic cells, and the downstream RAF/MEK/ERK signaling pathway and exerts anticancer effects by inhibiting proliferation, promoting apoptosis and reducing tumor angiogenesis.6,7 Studies have shown that the iron chelator deferoxamine significantly reduces the cytotoxic effect of sorafenib on an HCC cell line (Huh7), demonstrating that sorafenib exerts anticancer activity by inducing ferroptosis as a single agent in HCC cells, rather than through multikinase inhibitory effects.27,28 Moreover, in some HCC patients treated with sorafenib, concentration of serum oxidative stress response markers was associated with progression-free survival, suggesting that sorafenib-induced ferroptosis plays an important role in patient survival.41 Sorafenib has been recognized as an inducer of ferroptosis, which blocks cystine uptake by inhibiting system Xc- activity, reduces GSH biosynthesis, and thus induces ferroptosis and promotes oxidative stress to induce ferroptosis by increasing mitochondrial ROS production. The mechanism of action depends on the retinoblastoma status of HCC cells.27,28,42 A recent study suggested that sorafenib does not qualify as a bona fide ferroptosis inducer and does not induce ferroptosis in a range of tumor cell lines, in contrast to the cognate system Xc- inhibitors sulfasalazine and erastin.43 Interestingly, sorafenib both induces ferroptosis and protects HCC cells from erastin-induced ferroptosis by increasing the availability of amino acids for GSH synthesis through inhibition of protein biosynthesis.27,28,44 The existence of modes of antagonism may explain the apparent paradox, although further research is needed to explain it. Although sorafenib-induced ferroptosis also occurs in melanoma, pancreatic cancer, and colon cancer, both pharmacological inhibition (ferroptosis inhibitors) and genetic interference (RNA interference techniques) readily inhibit the antitumor effectiveness of sorafenib.27 Therefore, the precise mechanism of the ferroptosis-inducing effect of sorafenib needs further study to improve its cytotoxicity and weaken the drug resistance of patients. Nuclear factor E2-related factor 2 (NRF2) is a key regulator of antioxidative and electrophilic stress,45 and the activation and inactivation of NRF2 influences sorafenib-induced ferroptosis in HCC. For example, quiescin sulfhydryl oxidase 1 interacts with epidermal growth factor receptor to enhance its ligand-induced endosomal transfer and lysosomal degradation, resulting in moderation of NRF2 activation, disrupting redox homeostasis, and sensitizing HCC cells to oxidative stress, thereby enhancing sorafenib-induced ferroptosis.46 Glutathione S-transferase zeta 1 expression was found to be significantly decreased in sorafenib-resistant HCC cells, and downregulation of glutathione s-transferase zeta 1 inhibited sorafenib-induced HCC cell ferroptosis by increasing the levels of NRF2 and ferroptosis-related genes (GPX4 and SLC7A11).47 With sorafenib and erastin, p62/sequestosome-1 mediates the inactivation of Kelch-like ECH-associated protein 1 to prevent NRF2 degradation and enhance subsequent nuclear accumulation. NRF2 interacts with MAF bZIP transcription factor G and activates the transcription of genes involved in ROS and iron metabolism to confer ferroptosis resistance to HCC cells. Genetic or pharmacological inhibition of NRF2 expression/activity in HCC cells increases the anticancer activity of erastin and sorafenib in vitro and in vivo, indicating that the p62-Keap1-NRF2 antioxidant signaling pathway is a key negative regulator of ferroptosis in HCC cells.15 Another study showed that disulfiram (DSF)/Cu significantly impaired mitochondrial homeostasis, increased the free iron pool, and enhanced lipid peroxidation, leading to ferroptotic cell death. Inhibition of NRF2 expression via RNA interference or pharmacological inhibitors significantly facilitated the accumulation of lipid peroxidation, and rendered HCC cells more sensitive to DSF/Cu induced ferroptosis, which facilitated the synergistic cytotoxicity of DSF/Cu and sorafenib.48 Activation of NRF2 is essential for upregulation of metallothionein-1G (MT-1G) expression following sorafenib treatment, and MT-1G facilitates sorafenib resistance through inhibition of ferroptosis. Consequently, inhibition of MT-1G during therapy may be an option for HCC treatment.49 Sorafenib also induces ATP-binding cassette subfamily C member 5 expression through the phosphatidylinositol-3-kinase/AKT/NRF2 signaling pathway. Accumulation of ATP-binding cassette subfamily C member 5 increases intracellular GSH by interacting with and stabilizing SLC7A11, thereby attenuating lipid peroxidation and inhibiting ferroptosis.50 Shan et al.51 found that ubiquitin-like modifier activating enzyme 1 regulated the HCC cell phenotype and ferroptosis through the NRF2 signaling pathway to participate in the development of HCC.51 Thus, NRF2 is an important target of the ferroptosis network in HCC cells. Sigma 1 receptor (S1R) is a protein regulator associated with oxidative stress metabolism.52 Knockdown of S1R increases GSH depletion and inhibits the expression of ferritin heavy chain 1 and transferrin receptor protein 1 to promote iron enrichment and ROS accumulation to exert the anticancer activity of sorafenib.53 Haloperidol, an S1R antagonist, may benefit sorafenib-treated HCC patients by reducing the dose of sorafenib or enhancing the drug’s effectiveness.54 In conclusion, S1R protects HCC cells against sorafenib-induced ferroptosis. A better understanding of the role of S1R in ferroptosis may provide new insights into HCC treatment. The existence of sorafenib resistance is an important factor limiting the survival benefit of patients with advanced HCC. Yes-associated protein (YAP)/transcriptional coactivator with PDZ-binding motif (TAZ) induces the expression of SLC7A11 in a transcriptional enhanced associate domain (TEAD)-dependent manner. Moreover, YAP/TAZ maintains activating transcription factor 4 protein stability and transcriptional activity by limiting its polyubiquitination, which in turn synergistically induces SLC7A11 expression, thereby making HCC cells resistant to sorafenib-induced ferroptosis. The results indicate that the transcription factor YAP/TAZ is a key driver of sorafenib resistance in HCC.55 Other studies have demonstrated that YAP sensitizes HCC cells to ferroptosis by transcriptionally upregulating arachidonate lipoxygenase 3 leading to an increase of lipid peroxidation,56 and O-GlcNAcylation increases the ferroptosis sensitivity of HCC cells through the YAP/transferrin receptor pathway,57 suggesting that YAP may be an effective biomarker for predicting the response of HCC cells to ferroptotic induction. Deletion of azoospermia-associated protein 1 interferes with the ferroptotic effect of sorafenib on HCC cells by affecting the SLC7A11/GPX4 pathway.58 MiR-23a-3p is upregulated in sorafenib-resistant cells and recognizes and binds to acyl-CoA synthetase long-chain family member 4 (ACSL4) to limit accumulation of chelatable iron and negatively regulate sorafenib-induced HCC cell ferroptosis.59 Genetic and pharmacological inhibition of ACSL4 rescued sorafenib-induced ferroptosis in HCC cells in vitro and xenograft growth inhibition in vivo, suggesting that ACSL4 expression is a good candidate biomarker for predicting the ferroptotic sensitivity of HCC cells.60 In vivo and in vitro experiments have shown that sorafenib activates family with sequence similarity 134 member B-mediated autophagy of the endoplasmic reticulum in HCC, and targeting family with sequence similarity 134 member B-mediated reticulophagy activates sorafenib-induced ferroptosis in HCC cells.61 Deletion of LIFR promotes HCC tumorigenesis and confers resistance to sorafenib-induced ferroptosis. Mechanistically, loss of LIFR activates nuclear factor-κB signaling through Src homology region 2 domain-containing phosphatase-1, leading to upregulation of lipocalin 2 iron-chelating cytokine, thereby depleting intracellular iron to confer ferroptosis resistance in HCC cells.62 Upregulation of the secreted protein acidic and rich in cysteine promotes lactate dehydrogenase release and ROS production in HCC cells, sensitizing HCC cells to sorafenib.63 CDGSH iron sulfur domain 2 knockout promotes uncontrolled autophagy in sorafenib-resistant HCC cells to restore sorafenib-induced HCC cell ferroptosis and reverse drug resistance.64 CIARS (hsa_circ_0008367) was shown to be a promoter of ferroptosis in HCC cells after sorafenib treatment, in part due to the activation of human AlkB homolog H5-mediated autophagy and ferritin phagocytosis.65 Overall, the mechanism of sorafenib resistance is regulated by numerous complex gene networks, and the regulation of expression of the identified resistance-related genes during sorafenib treatment may be effective in improving it clinical benefit.
Research on noncoding RNAs (ncRNAs) is increasing, and many studies have shown that ncRNAs such as microRNAs, long noncoding RNAs (lncRNAs), and circular RNAs are key regulators in the tumorigenesis and development of HCC.66–68 However, whether ncRNAs mediate HCC cell ferroptosis remains unclear. Bai et al.69 demonstrated that miR-214 enhanced erastin-induced HCC cell ferroptosis by targeting activating transcription factor 4 expression in HCC cells. Exposure to erastin, the upregulated lncRNA GABPB1-AS1 in HCC inhibits peroxiredoxin-5 by blocking GABPB1 translation, leading to suppression of the cellular antioxidant capacity and cell viability.70 Peroxidases are key cellular antioxidant enzymes that block the destructive process of peroxidative damage to membranes caused by accumulated hydroxyl radicals. Xu et al.71 showed that circIL4R acts as an oncogene and relieves the inhibitory effect of miR-541-3p on GPX4 expression by sponging miR-541-3p, thereby upregulating GPX4 expression to inhibit ferroptosis in HCC cells. Similarly, Lyu et al.72 found that the circ0097009/miR-1261/SLC7A11 axis mediates HCC progression by regulating ferroptosis. Collectively, the studies demonstrated a mechanistic link between ncRNAs and the ferroptotic response in HCC that helps to elucidate the underlying mechanisms of ferroptosis in HCC cells, establishing ncRNAs as attractive therapeutic targets for HCC.
Ferroptosis is closely related to the development and treatment response of various cancers, and the inhibition of tumor progression by ferroptosis-inducing therapy may serve as a new therapeutic strategy for HCC.16 Sorafenib-induced ferroptosis promotes the antitumor mechanism of sorafenib, and ncRNAs have been shown to mediate HCC cell ferroptosis. In addition, other molecular targets that regulate ferroptosis in HCC have been described. Table 2 summarizes some molecular targets of ferroptosis in HCC.15,17,27,28,42,47–51,53–56,59,61–65,69–85 Figure 2 shows the regulatory pathways and some of the targets of ferroptosis in HCC. A previous report showed that YAP/TAZ promotes HCC cells to overcome sorafenib-induced ferroptosis in a TEAD-dependent manner.55 Integrative bioinformatics and experimental analysis revealed that TEAD can serve as a novel prognostic target for HCC, and that knockdown of TEAD2 induces ferroptosis through iron accumulation and subsequent oxidative damage.73 P53, a tumor suppressor implicated in the cell cycle, apoptosis, and cell senescence, has been reported to promote ferroptosis by inhibiting SLC7A11 expression and cystine uptake,74 whereas mutant p53 has been shown to inhibit the ferroptotic capacity of cells.86 Furthermore, zinc finger protein 498 inhibits the transcriptional activity of p53 by interfering with p53 Ser46 phosphorylation, thereby inhibiting apoptosis and ferroptosis in HCC cells.87 p53 has also been reported to negatively regulate ferroptosis, p53 limits erastin-induced ferroptosis in colorectal cancer cells by promoting the nuclear localization of dipeptidyl peptidase 4 and increasing the expression of SLC7A11.88 The above studies revealed that the dual effect of p53 on ferroptosis may depend on the cell type, a pivotal finding that provides the basis for the development of ferroptosis-inducing therapies based on p53-dependent tumor suppression. Tumor cells alter their susceptibility to ferroptosis through various gene expression and regulatory mechanisms, such as the increased glucose-6-phosphate dehydrogenase expression in HCC cells leading to an antiferroptotic state by inhibiting cytochrome P450 oxidoreductase expression.75 HCC cells with low H-ferritin expression are more sensitive to ras-selective lethal small molecule 3-induced ferroptosis.31 Xue et al.76 found that Rb1-inducible coiled-coil 1-induced upregulation of coiled-coil helix coiled-coil helix domain-containing protein 3 expression to stimulate mitochondrial function and increase ROS production, resulting in increased sensitivity of HCC cells to ferroptosis, and suggesting that Rb1-inducible coiled-coil 1 is a promising target for ferroptosis-based antitumor therapy. The occurrence of ferroptosis is accompanied by obvious morphological changes in mitochondria, and mitochondria are the main source of ROS,8,26 study of the regulators of mitochondrial function will help to elucidate the regulatory mechanism of ferroptosis. It has been reported that the RNA-binding protein alpha-enolase inhibited mitochondrial respiration through the alpha-enolase/iron regulatory protein 1/mitoferrin-1 axis, resulting in mitochondrial iron enrichment and excess accumulation of ROS in HCC cells.77 CDGSH iron sulfur domain 1 (CISD1), is a mitochondrial outer membrane protein that regulates mitochondrial iron uptake and respiration.89 Genetic inhibition of CISD1 potentiates erastin-induced ferroptosis by enhancing mitochondrial function, suggesting that CISD1 may be a negative regulator of ferroptosis in HCC cells.78 Ferritin heavy chain may serve as an intervention target for ferroptosis-inducing therapy, as it promotes ferroptosis resistance in HCC cells by regulating iron metabolism and maintaining mitochondrial homeostasis.90 Knockdown of isocitrate dehydrogenase 2 reduces the level of nicotinamide adenine dinucleotide phosphate, which is a key factor in maintaining the GSH-dependent mitochondrial antioxidant defense system, thereby promoting erastin-induced ferroptosis in HCC cells.79 Although the above reports have revealed the important role of mitochondrial homeostasis in ferroptosis, one study showed that mitochondria-deficient cells can still undergo ferroptosis.91 Therefore, the specific role of mitochondria in HCC cell ferroptosis needs further exploration. Normally differentiated cells mainly rely on mitochondrial oxidative phosphorylation for energy, whereas most tumor cells metabolize glucose into lactate by glycolysis under aerobic conditions to allow proliferation and invasion of cancer cells.92 High concentrations of lactate produced by aerobic glycolysis in HCC cells activates hydroxycarboxylic acid receptor 1 receptors on the plasma membrane and enhances monocarboxylate transporter 1-mediated lactate uptake, promotes ATP production, and deactivates the energy sensor, AMP-activated protein kinase (AMPK), resulting in upregulation of the expression of sterol regulatory element-binding protein 1 (SREBP1) and its target stearoyl-coenzyme A desaturase-1 (SCD1) to increase the production of anti-ferroptotic monounsaturated fatty acids. Additionally, high intracellular lactate concentrations inhibit ACSL4-mediated promotion of ferroptosis, which may synergize with the hydroxycarboxylic acid receptor 1/monocarboxylate transporter 1/AMPK/SREBP1/SCD1 axis-regulated anti-ferroptotic effect to amplify resistance to ferroptotic damage induced by common ferroptosis inducers such as ras-selective lethal small molecule 3 and erastin. The discovery of a lactate-mediated ferroptosis regulatory mechanism highlights its translational potential as a therapeutic target for ferroptosis-based tumor treatment.80 Notably, the AMPK/SREBP1 signaling pathway also influences the transcription of branched-chain amino acid transaminase 2. As a key regulator of glutamate metabolism, branched-chain amino acid transaminase 2 activates system Xc-activity through ectopic expression, protecting HCC cells from ferroptosis in vitro and in vivo.81
Systemic therapy involves anticancer drugs such as conventional cytotoxic and targeted agents to prevent tumor progression by inducing the death of cancer cells. However, the existence of intrinsic and acquired resistance limits the efficacy of drugs to a certain extent. The emergence of ferroptosis-inducing therapeutic strategies would effectively improve the resistance of cancer cells to anticancer drugs, and ferroptosis inducers can synergize with traditional antitumor drugs to better inhibit tumor progression. Several drugs that are already in clinical use or have a strong potential for clinical translation are known to promote ferroptosis (Table 1). Sorafenib is the first multikinase inhibitor approved for the treatment of patients with unresectable HCC, and is first-line targeted drug for advanced HCC6,7,27,28,41 The mechanism of sorafenib in HCC is described in detail in the previous discussion of ferroptosis and sorafenib in HCC. Natural products are an important source of anticancer drugs, and some natural products contain components that can decrease the viability of HCC cells by inducing ferroptosis.93 For example, in the presence of ferrostatin-1, saponin formosanin C-induced ROS formation was reduced, and inhibition of HCC cell viability was attenuated, suggesting that saponin formosanin C may act as a novel ferroptosis inducer.93 In addition to its antimalarial activity, artemisinin and its derivatives have been further validated for their anticancer properties in vivo and in vitro. Artesunate not only induces apoptosis and also triggers ferroptosis in HCC cells by promoting ferritinophagy and increasing intracellular free iron. Notably, artesunate was found to significantly enhanced the inhibitory effect of low-dose sorafenib in HCC cell lines and nude mouse xenografts by promoting lysosomal activation, ferritin degradation, lipid peroxidation, and subsequent sequential responses, including ferroptosis.35 Dihydroartemisinin induced ferroptosis by promoting the formation of phosphatidylethanolamine-binding protein 1/15-lipoxygenases and promoting cell membrane lipid peroxidation, thereby exerting anti-HCC activity.94 Of course, the cytotoxicity of artemisinin and dihydroartemisinin in normal cells is not negligible, and it remains to be determined whether the minimal toxicity profile observed in clinical trials can enable their cell death-promoting activity to be utilized. Heteronemin also induces both apoptosis and ferroptosis of HCC cells, but its cytotoxicity limits its use. Hepatic arterial infusion chemotherapy is a feasible strategy to deliver drugs directly to the tumor and minimize systemic toxicity.95 Another study showed that solanine promoted ferroptosis in HCC cells via GPX4-induced disruption of the GSH redox system.32 In addition, the Chinese medicine atractylodin induced ferroptosis in HCC cells by inhibiting GPX4 expression and upregulating ACSL4 expression.33 In conclusion, the above natural products may be used as ferroptosis inducers to exert anticancer effects, but further clinical trials are needed to verify their efficacy and explore their side effects. Increased GSH depletion via cysteine deprivation or cysteinase inhibition enhances oxidative stress and mitochondrial ROS accumulation, leading to lipid peroxide overload and ferroptosis. Therefore, modulating extracellular cysteine levels may open up new therapeutic options for ferroptosis-inducing cancer therapy, especially in combination with ROS-inducing drugs, such as synergistic cysteine depletion with sorafenib, increasing the susceptibility of HCC cells to ferroptosis.96 There are also some drugs that lead to HCC cell ferroptosis by inducing strong mitochondrial dysfunction to induce ROS overproduction combined with the breakdown of the antioxidant defense system, such as the nuclear protein 1 inhibitor ZZW-115.97
Radiotherapy benefits patients with unresectable or advanced HCC, but its effectiveness is hampered by radioresistance and side effects.98 Radiotherapy has been reported to induce ferroptosis in cancer cells, including HCC, fibrosarcoma, and breast cancer cells, by inhibiting SLC7A11 or activating ACSL4 expression, thereby increasing lipid peroxidation and subsequent oxidative damage.17,99 In addition, ferroptosis inducers enhances the antitumor activity of radiation in selected xenograft tumor models,100 suggesting that induction of ferroptosis may be an option to overcome radioresistance. Studies have shown that collectrin, regulated by the nuclear respiratory factor 1/Ran (Ras-related nuclear protein)/dihydrolipoamide dehydrogenase protein complex, acts as a radiation target and enhances the radiosensitivity of HCC cells by inducing ferroptosis.17 Low expression of the copper metabolic gene MURR1 domain 10 induced by ionizing radiation promotes hypoxia-inducible factor-1α (HIF1α) nuclear translocation and transcription of ceruloplasmin (CP) and SLC7A11 by inducing intracellular Cu accumulation, which jointly inhibits HCC ferroptosis leading to radioresistance. In addition, elevated CP enhances the expression of HIF-1α, forming a positive feedback loop of HIF1α/CP, thereby promoting HCC radioresistance.82 CP inhibits ferroptosis by maintaining Cu-Fe homeostasis in HCC cells (Fig. 3).101 Gene sensitization therapy effectively improves the efficacy of radiotherapy while reducing the dose of radiotherapy and the side effects of radiotherapy. Radiotherapy synergistically overexpressing collectrin and copper metabolic gene MURR1 domain 10 may effectively improve the survival benefit of HCC patients.
The advent of immunotherapy based on immune checkpoint inhibitors (ICIs) targeting programmed cell death protein 1 and its ligands programmed death-ligand 1 and cytotoxic T lymphocyte-associated protein 4 has improved the prognosis of HCC patients.102 Immunotherapy mainly exerts anticancer effects by activating potent cytotoxic T cell-driven antigen-antibody responses. Studies have revealed that activated cytotoxic T cells under immunotherapy downregulate the expression of solute carrier family 3 member 2 and SLC7A11 by releasing interferon-gamma-γ to activate the Janus kinase/signal transducer and activator of transcription signaling pathway to inhibit system Xc- activity and promote mitochondrial damage-related lipid peroxidation, thereby inducing HCC cell ferroptosis.83 Moreover, ferroptotic HCC cells release tumor-associated antigens that improve the immunogenicity of the tumor microenvironment and enhance the effect of immunotherapy.18,83 Furthermore, anti-programmed death-ligand 1 monoclonal antibodies and ferroptosis inducers synergistically inhibit tumor growth in vitro and in vivo.18 Transforming growth factor β1 released by macrophages induces HCC cell ferroptosis by activating the small mother against decapentaplegic to inhibit xCT (the catalytic subunit of the system Xc-) expression in a time- and dose-dependent manner.84 Some immune cells in the tumor microenvironment, such as monocytes, macrophages, and natural killer cells, may be involved in the maintenance of iron homeostasis.103 Therefore, ferroptosis is expected to provide new ideas and targets for improving the efficacy of immunotherapy in HCC patients (Fig. 3). In addition, signatures constructed and validated based on ferroptosis- and immune-related genes can effectively predict the expression levels of ICI-related targets, and provide insights into the selection and efficacy prediction of ICIs in HCC patients.104,105
Interventional therapy, which causes ischemic necrosis of tumor tissue by blocking the blood supply to the tumor, has been used as a safe and effective treatment option for unresectable HCC.106 However, the activation of HIF-1α and vascular endothelial growth factor in the hypoxic microenvironment by transcatheter arterial chemoembolization increases the potential of tumors for angiogenesis, recurrence, and metastasis.107,108 Hypoxia induced by interventional embolization treatment inhibits methyltransferase-like 14 (METTL14) expression in a HIF-1α-dependent manner, thereby blocking METTL14/YTH domain family 2/SLC7A11 axis-mediated ferroptosis and promoting HCC progression. These investigations highlight hypoxia-regulated ferroptosis in HCC cells and identify the HIF-1α/METTL14/ YTH domain family 2/SLC7A11 axis as a potential therapeutic target for HCC interventional embolization treatment (Fig. 3).108
Using nanoparticles to deliver and control drug release has unique advantages, such as high drug loading, targeting of specific tissues and organs, and improved pharmacokinetic properties.109 Constructing efficient ferroptosis- related nanoplatforms and developing novel ferroptosis inducers will increase the efficacy of existing ferroptosis inducers and make full use of existing clinical anti-HCC drugs to expand their therapeutic value (Fig. 3). For example, a novel cascaded copper-based metal-organic framework therapeutic nanocatalyst developed by Tian et al.110 activates ferroptosis by GSH depletion and promotes substantial accumulation of lipid peroxidation, resulting in cascade-amplified ferroptosis mediated HCC therapy. Low-density lipoprotein docosahexaenoic acid nanoparticles have been shown to induce ferroptotic cell death in HCC by pronounced lipid peroxidation, depletion of GSH and inactivation of the lipid antioxidant GPX4.111 Nanoparticles loaded with ferroptosis inducers have also been studied. Tang et al.112 reported that spontaneous degradation of manganese-oxygen bonds in sorafenib-loaded manganese-doped mesoporous silica nanoparticles together with the on-demand release of sorafenib achieved dual depletion of GSH and blocked its synthesis, thereby exerting an efficient tumor suppressor effect. Similarly, the antitumor activity of sorafenib was enhanced upon delivery of sorafenib-loaded MIL-101(Fe) nanoparticles.113 Combination therapy can overcome the shortcomings of monotherapy and reduce resistance induced by a single agent.114 Nanoplatforms incorporating ferroptosis inducers, chemotherapy drugs, or other therapies such as sonodynamic therapy can effectively inhibit HCC progression. Zhou et al.115 designed a multifunctional nanoplatform with the potential to integrate cancer diagnosis, treatment, and monitoring. It provided a novel clinical antitumor therapeutic strategy to induce ferroptosis via the consumption of GSH, disrupt redox balance by the Fenton reaction and doxorubicin-supplied hydrogen peroxide, synergistic cytotoxicity of doxorubicin inhibition of recurrence and metastasis of HCC, and reversing drug resistance in translational therapy. Chen et al.116 assessed the prospect of nanobubbles combined with SDT and ferroptosis for treating HCC. Because of their low immunogenicity, low cytotoxicity, and high biocompatibility, nanosized vesicle exosomes can be used as drug delivery systems.117 Du et al.118 designed engineered exosomes composed of CD47, erastin, and rose bengal that delivered erastin and rose bengal to tumor tissues with high specificity. They avoided phagocytosis by the mononuclear phagocyte system and achieved high distribution in tumor tissues, thereby inducing intensive ferroptosis in HCC with minimal liver and kidney toxicity.
Clinical management emphasizes early and effective disease screening, diagnosis, treatment, and prognostic prediction, so the development of biomarkers for tumor detection and diagnosis is extremely important. Accumulating evidence suggests that ferroptosis is involved in cancer development and treatment response, and the identification of ferroptosis-related genes and pathways will provide references for the clinical management of HCC patients. Investigators used comprehensive bioinformatics analysis to screen out genes associated with HCC and ferroptosis, such as ubiquitin-like modifier activating enzyme 1,51 heat shock protein beta-1,119 six-transmembrane epithelial antigen of the prostate family member 3,120 ATP-binding cassette transporter 6 of subfamily B,19 cysteine-preferring transporter 2,121 SLC7A11,122 and some lncRNAs.123,124 The differential expression of those genes may be associated with poor prognosis of HCC patients and they may also serve as effective biomarkers for the diagnosis of HCC patients. The tumor node metastasis classification system has been widely used to predict prognosis and guide treatment in clinical practice, but patients at the same stage may have different prognoses. Therefore, generation of an accurate, safe, and effective novel prognostic model is needed to assist and supplement the tumor node metastasis classification system. Prognostic models constructed and validated by some teams based on ferroptosis-related genes can predict the OS of HCC patients.125–127 However, the heterogeneity of HCC and the complexity of the tumor microenvironment limit the specificity and sensitivity of ferroptosis-related gene-based prognostic models. In another study, prognostic models constructed by combining ferroptosis-related genes with immune-related genes,104,105 pyroptosis-related genes,128 and hypoxia-related genes,129–130 accurately predicted patient prognosis and provided a reference for the selection and prediction of the efficacy of immune checkpoint inhibitor therapy for HCC patients. Of course, the prognostic models are constructed based on retrospective data analysis, lack validation by multicenter prospective cohort studies, and have certain biases. More in vitro and in vivo studies are required for further verification.
Iron overload and oxidative stress are important triggers of most liver diseases,16 ferroptosis can affect various liver diseases including ischemia/reperfusion-related injury,16 alcoholic liver disease,131 and nonalcoholic fatty liver disease132 by regulating the levels of iron and ROS. Interestingly, inhibition of ferroptosis can counteract the pathophysiological progression of liver diseases such as alcoholic liver disease and nonalcoholic fatty liver disease.131,132 However, induction of ferroptosis enhances the sensitivity of HCC patients to sorafenib.55,58 The inflammatory reaction caused by long-term chronic liver injury and repair is conducive to liver fibrosis and even HCC. Therefore, it is crucial to explore the optimal time to intervene in ferroptosis to prevent the progression of chronic liver disease to HCC. Differences in the pathogenesis of liver diseases will determine differences in preferred treatment. The mechanism of ferroptosis in different liver diseases should be fully considered to determine its mode of action.
In recent years, major progress in research on ferroptosis in cancer prevention, diagnostics, prognostics, and treatment has been reported. This review briefly introduces ferroptosis-related regulatory pathways and typical ferroptosis inducers and inhibitors, and describes the regulatory mechanism by which sorafenib inhibits HCC progression by inducing ferroptosis. It describes new targets of ferroptosis in HCC cells and the use of ferroptosis in the treatment, diagnosis, and prognosis of HCC. The complex regulatory network of ferroptosis in HCC pathophysiology and therapy indicates that its clinical translational anticancer strategy may be laborious, and that further research is needed to optimize individualized and precise treatment strategies for HCC patients and improve survival benefits. In addition, the differential susceptibility of HCC patients with a complex genetic background to ferroptosis-inducing therapy will limit the individualized and precise treatment of HCC patients, so it is critical to identify biomarkers associated with responsiveness through liquid biopsy of blood, bile, and urine, and tumor tissue sample analysis to identify HCC patients who are sensitive to ferroptosis-inducing therapy. The following issues still to be resolved. First, the ferroptotic response is regulated by a complex network of epigenetic, transcriptional, post-transcriptional and post-translational mechanisms. Although some targets have been identified, further investigation of underlying signal transduction pathways and key transcriptional regulators of ferroptosis in HCC is needed. In addition to drug resistance, the role of ferroptosis in other malignant phenotypes of HCC such as invasion, metastasis, metabolism, and autophagy cannot be ignored. Second, the development of drugs that directly target ferroptotic pathways might provide new strategies for tumor treatment. Ferroptosis-induced therapy combined with other anticancer therapies, such as immunotherapy or radiotherapy effectively suppress tumor growth by inducing a mixed-type RCD, but the identification of ferroptosis inducers remains to be done in cell and animal experiments. Clinical use is still in the future. In addition, the side effects of ferroptosis inducers are still unclear, the effectiveness of the lowest dose of ferroptosis inducers has not yet been determined, and the effectiveness of sorafenib in clinical use is limited by drug resistance. Therefore, reducing the side effects of ferroptosis inducers or reversing drug resistance remains a challenge in clinical oncology. Of course, exploring the complex crosstalk between ferroptosis and other RCDs will provide a reference for translational medicine based on ferroptosis. Tackling the key issues discussed in this review will deepen our understanding of the significant role that ferroptosis plays in HCC, thus providing a new scientific basis for targeting ferroptosis to prevent and treat HCC. |
|
PMC9647099 | Marco Ferronato,Claudine Lalanne,Chiara Quarneti,Michele Cevolani,Chiara Ricci,Alessandro Granito,Luigi Muratori,Marco Lenzi | Paraneoplastic Anti-Tif1-gamma Autoantibody-positive Dermatomyositis as Clinical Presentation of Hepatocellular Carcinoma Recurrence | 24-05-2022 | Dermatomyositis,Anti-transcriptional intermediary factor-1 gamma (tif1-γ) antibodies,Hepatocellular carcinoma,Paraneoplastic syndrome | Hepatocellular carcinoma (HCC) is rarely associated with autoimmune paraneoplastic syndromes. We report a case of anti-transcriptional intermediary factor-1 gamma (TIF1-γ)-positive dermatomyositis (DM) as clinical presentation of HCC recurrence in a 72-year-old male patient admitted to our hospital due to fatigue, myalgia, and typical skin rash. His medical history was notable for hepatitis C-related cirrhosis, successful treatment with direct-acting antiviral agents, and previously efficacious treatment of HCC. Laboratory testing showed significant rhabdomyolysis with anti-TIF1-γ antibodies at high titer, and DM was diagnosed. After a careful diagnostic workup, HCC recurrence was diagnosed. After first-line corticosteroid treatment, azathioprine and intravenous immunoglobulin treatments were administered; unfortunately, he mounted only partial response. Owing to the compromised performance status, no HCC treatment was feasible, and, according to international guidelines, he received only best supportive care. Here, we discuss the diagnostic, prognostic, and pathogenic roles of anti-TIF1-γ antibodies associated with paraneoplastic DM and the scant literature data on its occurrence in HCC patients. Considering the TIF1 gene family’s established role in oncogenesis, we also review the role of TIF1-γ as a tumor-related neoantigen, leading to the development of clinically overt anti-TIF1-γ antibodies-positive DM. | Paraneoplastic Anti-Tif1-gamma Autoantibody-positive Dermatomyositis as Clinical Presentation of Hepatocellular Carcinoma Recurrence
Hepatocellular carcinoma (HCC) is rarely associated with autoimmune paraneoplastic syndromes. We report a case of anti-transcriptional intermediary factor-1 gamma (TIF1-γ)-positive dermatomyositis (DM) as clinical presentation of HCC recurrence in a 72-year-old male patient admitted to our hospital due to fatigue, myalgia, and typical skin rash. His medical history was notable for hepatitis C-related cirrhosis, successful treatment with direct-acting antiviral agents, and previously efficacious treatment of HCC. Laboratory testing showed significant rhabdomyolysis with anti-TIF1-γ antibodies at high titer, and DM was diagnosed. After a careful diagnostic workup, HCC recurrence was diagnosed. After first-line corticosteroid treatment, azathioprine and intravenous immunoglobulin treatments were administered; unfortunately, he mounted only partial response. Owing to the compromised performance status, no HCC treatment was feasible, and, according to international guidelines, he received only best supportive care. Here, we discuss the diagnostic, prognostic, and pathogenic roles of anti-TIF1-γ antibodies associated with paraneoplastic DM and the scant literature data on its occurrence in HCC patients. Considering the TIF1 gene family’s established role in oncogenesis, we also review the role of TIF1-γ as a tumor-related neoantigen, leading to the development of clinically overt anti-TIF1-γ antibodies-positive DM.
Hepatocellular carcinoma (HCC) is the fifth most common cancer and the third most frequent cause of cancer-related death worldwide, with more than 900,000 new cases and more than 800,000 deaths in 2020 alone. Many patients have no symptoms related to the tumor, especially among those who have been undergoing regular surveillance and in whom HCC is detected at an early stage.1 Limited data are available on the prevalence of paraneoplastic syndromes in HCC patients. However, the most frequently reported are hypercholesterolemia, hypoglycemia, hypercalcemia, and erythrocytosis.2–4 Other rarer conditions include erythema nodosum, polyarthritis, carcinoid syndrome, hemophagocytic syndrome, and porphyria cutanea tarda.5–8 It has been speculated that these syndromes might arise from tumor secretion of hormones, peptides, or cytokines, as well as immune cross-reactivity between malignant and normal tissues and are not directly related to the physical effects of the primary or metastasis tumors.5 The association between HCC and dermatomyositis (DM) has so far been sparsely reported (Table 1).9–20 We, herein, describe a case of a newly diagnosed and immunologically well-characterized paraneoplastic DM as the clinical presentation of recurrent HCC.
We report the case of a 72-years-old male patient admitted to our Internal Medicine and Immunorheumatology Unit in September 2020 due to severe fatigue, widespread myalgia, pain in his right shoulder exacerbated by movement, and itching. The patient’s medical history included hepatitis C virus (HCV)-related cirrhosis that had been successfully treated 5 years earlier with direct-acting antiviral agents (ledipasvir/sofosbuvir). After the antiviral treatment, owing to the cirrhotic stage, he continued the imaging surveillance program. In April 2016, an HCC nodule (1.8 cm) was diagnosed at segment IV and successfully treated with percutaneous radiofrequency ablation. In March 2020, during the follow-up, a small recurrent HCC nodule (1.5 cm) was diagnosed at segment VI and treated with trans-arterial chemoembolization (TACE). There was no imaging evidence of persistent/recurrent HCC at the scheduled 1-month and 3-month imaging follow-up nor during the regularly programmed 6-month imaging follow-up after TACE. Alpha-fetoprotein (AFP) serum levels during the follow-up were persistently normal. Physical examination revealed poikiloderma in photo-exposed areas of the face, upper back (“shawl sign”), and a rash on the anterior neck and upper chest (“V-sign”) associated with edema of the upper limbs and scratching injuries. The patient also presented symmetric and proximal muscle weakness, weak osteotendinous reflexes of the upper limbs, and muscle tenderness of the right shoulder. At the clinical presentation, the patient’s laboratory data were as follows: aspartate aminotransferase (AST) 493 U/L [upper normal level (UNL) 50 U/L]; alanine aminotransferase (ALT) 94 U/L (UNL <50 U/L); gamma-glutamyl transferase (γGT) 82 U/L (UNL <55 U/L); alkaline phosphatase 88 U/L (UNL <120 U/L); lactate dehydrogenase (LDH) 631 U/L (UNL 248 U/L), aldolase 59 U/L (UNL 7.6 U/L); creatinine kinase 10.906 U/L (UNL 170 U/L); and C-reactive protein 6.17 mg/dL (UNL 0.5 mg/dL). Findings for HCV RNA, HBsAg and HBV-DNA were negative. Findings for anti-nuclear antibodies (ANAs), anti-double-stranded-DNA, anti-U1RNP, anti-Sm, anti-SSA, anti-SSB, anti-Scl 70, anti-Jo1, anti-PCNA, anti-PM-Scl, anti-ribosomial-P-protein, anti-CENP-B, anti-fibrillarine, anti-RNA polymerase-III and anti-neutrophil cytoplasmic antibodies (ANCA) were negative. Of relevance, a comprehensive immunoblot assay (EUROLINE Autoimmune Inflammatory Myopathies DL 1530-5001-4 G; Euroimmun, Lübeck, Germany) for testing autoantibodies direct against a wide range of myositis-specific antigens (Mi-2α, Mi-2β, TIF1γ, MDA5, NXP2, SAE1, Ku, PM-Scl100, PM-Scl75, JO-1, SRP, PL-7, PL-12, EJ, OJ, Ro-52, cN-1A) revealed anti-TIF1γ antibody positivity at high titer, while the findings for other myositis-related autoantibodies were all negative. Skin biopsy showed thinned epidermis, vascular alteration at the dermo-epidermal junction, edema of papillar derma, perivascular lymphocytic infiltrate, and mucin deposition in the papillar and medium derma (Fig. 1). Electromyography revealed muscle-related features consistent with a myopathic pattern. According to the Bohan and Peter criteria (Table 2),20 DM was diagnosed and treatment with prednisone (1 mg/Kg) was started. A gradual decrease of serum muscle enzyme levels was observed after steroid treatment (Fig. 2). Re-testing for HCV RNA again produced negative results; to exclude a paraneoplastic origin of DM, a careful imaging workup was performed. In September 2020, a total body contrast-enhanced CT scan did not reveal cancer. In particular, no evidence of recurrent/residual HCC was detected. The AFP serum level was within the normal range. Muscle strength and skin lesions’ recovery was incomplete, and a few weeks later, the patient complained of dysphagia. One month after the start of immunosuppressive therapy, he was admitted again to our Unit for aspiration pneumonia and ascites. He was treated with piperacillin/tazobactam, albumin infusion, and furosemide. Fiberoptic Endoscopic Evaluation of Swallowing (FEES) showed complete dysphagia, so parenteral nutrition was started. For severe DM-related symptoms, immunosuppressive treatment with prednisone 1 mg/kg was continued, and treatment with intravenous immunoglobulin (IVIG) 0.4 g/kg/die for 5 days was begun. In October 2020, a contrast-enhanced ultrasound (CEUS) of the abdomen showed a focal liver lesion of segment II (13×15 mm) characterized by arterial phase hyperenhancement and wash-out in portal phase [CEUS-Liver Imaging Reporting and Data System (LI-RADS) LR-5] (Fig. 3A, B) and hilar metastatic lymphadenopathy (20 mm × 35 mm; Fig. 3C). According to the American College of Radiology and the international guidelines on diagnosis and management of HCC, a diagnosis of HCC was made.21–24 According to the Barcelona Clinic Liver Cancer Staging System (BCLC), the patient had an advanced HCC stage (BCLC-C).25 Treatment of HCC could not be performed because of the patient’s poor physical conditions (Performance Status 3 according to The Eastern Cooperative Oncology Group score). We continued immunosuppressive treatment with prednisone, gradually tapered, and initiated azathioprine 50 mg/day as a steroid-sparing immunosuppressive agent, along with a monthly infusion of IVIG (2 g/kg). At the last follow-up, after 12 months, the abdominal CT showed HCC intrahepatic (new lesions) and extrahepatic (enlargement of hepatic hilar lymph node and newly detected abdominal lymph nodes) disease progression (Fig. 4), with the patient’s clinical conditions deteriorating rapidly.
DM is an idiopathic inflammatory myopathy, featuring proximal skeletal muscle weakness and evidence of muscle inflammation associated with a characteristic rash. Cutaneous manifestations more commonly precede muscle weakness and may also develop in the absence of detectable muscle disease. DM is also associated with inflammatory arthritis, interstitial lung disease, the Raynaud phenomenon, and presence of autoantibodies.26 Population-based cohort studies from several countries and metanalyses have confirmed the increased risk of cancer among patients with inflammatory myopathies, mainly DM. In fact, up to one-third of cases are paraneoplastic.27–31 Malignancy can be diagnosed before, simultaneously to, or after the diagnosis of DM, and adenocarcinomas of the cervix, lung, ovaries, pancreas, bladder, and stomach account for approximately 70% of the cancers associated with inflammatory myopathies.32,33 On the contrary, the association of DM with HCC is infrequent, and there are only very few case reports describing HCC-related DM (Table 1). In all described patients, DM partially improved under corticosteroid treatment. However, each patient’s prognosis was poor due to tumor progression or recurrence. Of potential interest, in almost all cases of HCC-associated DM above-mentioned, patients had HBV and/or HCV as the etiology of underlying liver disease. According to this association, a DM induction linked to viral infection has been speculated.5,13 In our case, there was no evidence of HCV recurrence after 5 years of successful antiviral therapy. Further, given that autoimmune extrahepatic HCV-related manifestations are generally mitigated by antiviral treatment, we believe that in our patient DM can be considered as a pure paraneoplastic manifestation, in line with other paraneoplastic DM reported in the literature.34 Of interest, DM has never been previously immunologically characterized as anti-TIF1-γ antibody positive in HCC and, of clinical relevance, has never been described as the clinical presentation of tumor recurrence after prior proven curative treatment. Serum autoantibodies that confer a positive risk of malignancy are the ones directed to anti- TIF1-γ (anti-p155, anti-p155/140) and to nuclear matrix protein (NXP)-2 (anti-MJ or anti-p140), as they are present in 83% of patients with cancer-associated DM.35,36 In DM patients with antibodies to TIF1-γ, those harboring a malignancy range from 42% to 100%, depending on the study.36 Research into paraneoplastic syndromes has always been considered a good way to shed light on tumor development, maintenance, and proliferation. Human TIF1-γ, also known as ectodermin or tripartite motif (TRIM) 33, is encoded by the TRIM33 gene, located on the p13 band of the short arm of chromosome 1. TIF1-γ has been shown to serve as a transcription regulator, a tumor suppressor, a mediator of DNA damage repair, and an E3 ligase that modulates TGF-β signaling. However, mechanisms leading to anti-TIF1-γ autoantibody appearance and the DM development in patients with cancer have not been firmly established.37,38 Ding et al.37 demonstrated the dual role of TIF1-γ in the HCC oncogenesis, which relies on the role of TGF-β under different cellular contexts; for example, in the early stage of HCC, TIF1-γ promotes cell growth by reducing the cytostatic effect of TGF-β, while in the advanced stage of HCC, TIF1-γ acts as a tumor suppressor gene as it inhibits TGF-β-induced epithelial-mesenchymal transition and TGF-β/Smad downstream metastatic signaling. Importantly, HCC patients with low TIF1-γ expression had shorter overall survival and higher recurrence rates than those with high TIF1-γ expression. TIF1-γ down-regulation was correlated with loss of tumor encapsulation, vascular invasion, malignant differentiation, and a more advanced BCLC stage.37 In TIF1-γ cancer-associated DM, it has been hypothesized that cancer is the underlying driver of DM and that TIF1-γ may function as a tumor autoantigen. Recent findings support the hypothesis that tumors from paraneoplastic anti-TIF1-γ-positive patients show increased genetic alterations, such as mutations and loss of heterozygosity in TIF1 genes. In the context of a high expression of TIF1-γ in the patients’ tumor, muscle, and skin, these genetic alterations may be key to understanding the genesis of paraneoplastic myositis. Therefore, aberrant TIF1-γ in cancer cells might develop neoantigens targeted by the immune response, which cross-react with the wild-type skin and muscle antigens (Fig. 5).38,39 The latter can be effective by eradicating tumor cells without clinical symptoms or unsuccessful with tumor development. The anti-tumor immune response cross-reacts with wild-type TIF1-γ in muscle and skin cells. The high content of TIF1-γ antigens in muscle and skin might explain why the anti-tumor response leads to the clinical manifestation of DM. However, a genetic predisposition could also explain why not all tumors with a dysregulated expression of TIF1-γ lead to DM. It has been previously reported that myositis is associated with specific HLA sequences (HLA-DRB1*0301 and HLA-DQA1*0501 in Caucasians), and TIF1-γ autoantibodies have been associated with HLA-DQA1*0301. In addition, a significant HLA association was observed in patients with TIF1-γ autoantibodies with the HLA DQB1*02 allele group.34 From an overall perspective, we can hypothesize a tumor-induced autoimmune pathophysiology of TIF1-γ paraneoplastic DM, as shown in Figure 5. It is therefore conceivable that at the time of first occurrence (April 2016) and recurrence (March 2020) of HCC, the TIF1 mutation was not yet present, whereas its later acquisition led to the clinical onset of DM (September 2020) with subsequent imaging evidence of tumor recurrence (October 2020). Of clinical relevance, a negative prognostic impact of paraneoplastic DM has been previously reported, as our case would confirm.2,3 In conclusion, we describe for the first time a case of paraneoplastic anti-TIF1-γ antibody-positive DM as a clinical presentation of recurrent HCC. The development of tumor-induced autoantibodies further evidences the prominent pathogenic role and the complexity of immunological mechanisms of the tumor microenvironment driving tumor growth and spread. |
|
PMC9647103 | Sheng-Liang Xin,Xiao-Li Pan,Xiao-Yuan Xu,Yan-Yan Yu | USP10 Alleviates Palmitic Acid-induced Steatosis through Autophagy in HepG2 Cells | 31-03-2022 | Autophagy,Nonalcoholic fatty liver disease,Steatosis,Ubiquitin-specific peptidase-10,C-jun N-terminal protein kinase-1,Tuberous sclerosis complex-2 | Background and Aims Nonalcoholic fatty liver disease (NAFLD) is a common chronic liver disease caused by over-nutrition. Impaired autophagy is closely related to NAFLD progression. Recently, ubiquitin-specific peptidase-10 (USP10) was reported to ameliorate hepatic steatosis, but the underlying mechanism is still unclear. In view of the potential effects of USP10 on autophagy, we investigated whether USP10 alleviated steatosis through autophagy. Methods HepG2 cells were treated with palmitic acid (PA) to model NAFLD in vitro. Lentivirus was used to regulate USP10 level in cells. Autophagic regulators were used to autophagic progression in cells. Western blotting, real-time fluorescence quantitative polymerase chain reaction, lipid drop staining and immunofluorescent staining were performed to determine the effect of USP10 on lipid autophagy. Student’s t-test and Tukey’s post hoc test were used to compare the means among groups. Results PA induced cellular steatosis with dependance on autophagy. USP10 overexpression alleviated PA-induced steatosis, restored autophagic activity, promoted autophagic flux, including synthesis and degradation of autophagosomes, and lipid-targeted autophagy. In the presence of autophagy inhibitors, the protective effectiveness of USP10 on steatosis decreased. Furthermore, the specific inhibitor to C-jun N-terminal protein kinase-1 (JNK1), DB07268, abolished USP10-induced autophagy. However, during early stage inhibition of JNK1, compensatory expression of tuberous sclerosis complex-2 (TSC2) maintained autophagy. The degree of TSC2-to-JNK1 compensation was positively associated with USP10 level. Functionally, JNK1 and TSC2 were involved in the lipid-lowering effect of USP10. Conclusions USP10 alleviated hepatocellular steatosis in autophagy-dependent manner. JNK1/TSC2 signaling pathways were required for USP10-induced autophagy. | USP10 Alleviates Palmitic Acid-induced Steatosis through Autophagy in HepG2 Cells
Nonalcoholic fatty liver disease (NAFLD) is a common chronic liver disease caused by over-nutrition. Impaired autophagy is closely related to NAFLD progression. Recently, ubiquitin-specific peptidase-10 (USP10) was reported to ameliorate hepatic steatosis, but the underlying mechanism is still unclear. In view of the potential effects of USP10 on autophagy, we investigated whether USP10 alleviated steatosis through autophagy.
HepG2 cells were treated with palmitic acid (PA) to model NAFLD in vitro. Lentivirus was used to regulate USP10 level in cells. Autophagic regulators were used to autophagic progression in cells. Western blotting, real-time fluorescence quantitative polymerase chain reaction, lipid drop staining and immunofluorescent staining were performed to determine the effect of USP10 on lipid autophagy. Student’s t-test and Tukey’s post hoc test were used to compare the means among groups.
PA induced cellular steatosis with dependance on autophagy. USP10 overexpression alleviated PA-induced steatosis, restored autophagic activity, promoted autophagic flux, including synthesis and degradation of autophagosomes, and lipid-targeted autophagy. In the presence of autophagy inhibitors, the protective effectiveness of USP10 on steatosis decreased. Furthermore, the specific inhibitor to C-jun N-terminal protein kinase-1 (JNK1), DB07268, abolished USP10-induced autophagy. However, during early stage inhibition of JNK1, compensatory expression of tuberous sclerosis complex-2 (TSC2) maintained autophagy. The degree of TSC2-to-JNK1 compensation was positively associated with USP10 level. Functionally, JNK1 and TSC2 were involved in the lipid-lowering effect of USP10.
USP10 alleviated hepatocellular steatosis in autophagy-dependent manner. JNK1/TSC2 signaling pathways were required for USP10-induced autophagy.
Nonalcoholic fatty liver disease (NAFLD) is the most common chronic liver disease worldwide, including developing countries. NAFLD is a metabolic disease, with significant accumulation of lipid droplets (LDs) in hepatocytes that is usually associated with oxidative stress, inflammation, fibrogenesis, and insulin resistance. Progressive NAFLD can develop into nonalcoholic steatohepatitis (NASH) or hepatocellular carcinoma. NAFLD accompanied with obesity, type 2 diabetes mellitus, and cardiovascular events increases overall mortality.1 NAFLD has become the most frequent indication for liver transplantation.2 An international expert consensus recommended renaming NAFLD as metabolic dysfunction-associated fatty liver disease.3 The pathogenesis of NAFLD is unclear and the treatment of NAFLD is limited. Autophagy is a highly conserved intracellular activity that degrades cellular components, such as impaired organelles and redundant metabolites. The classic autophagy process starts with the synthesis of autophagosomes. Microtubule-associated protein light chain (LC)3B-1, the inactive form of LC3B, located in cytosol, is recruited to membranes of autophagosomes and convert to its active form, LC3B-2. Target substances are engulfed by autophagosomes, which then fuse with lysosomes to form autolysosomes, in which the engulfed substances are degraded.4 Autophagy is regulated by a cluster of autophagy target genes (ATGs) and is induced by energic stress and messengers like AMP-activated protein kinase (AMPK).5 Lack of ATG56 or ATG77 and autophagic inhibitors6 increases the concentration of triglyceride in hepatocytes. Palmitate acid (PA) has been shown to suppress autophagy in hepatocytes in a time-dependent manner.8 Some compounds that promote autophagy alleviate hepatic steatosis.9,10 Therefore, autophagy plays a key role in NAFLD progression. Ubiquitin-specific peptidase (USP)-10 is a deubiquitinating enzyme located in both the cytoplasm and nucleus,11 and its therapeutic effects on oxidative stress, infection, heat shock,12,13 cardiac hypertrophy,14 and hepatic ischemic/reperfusion injury15 have been studied. The protective effects of USP10 in NAFLD, include inhibition of hepatic steatosis, insulin resistance, and inflammation via sirtuin-6.16 USP10 is regulated by the long noncoding RNA Mirt2 and micro (mi)RNA-34a-5p to suppress hepatic steatosis.17 USP10 interacts with Beclin1, a member of the autophagic complex, to stabilize p53 and inhibit cell proliferation,18 and LC3B ubiquitination is reversed in H4 cells by USP10.19 However, the direct role of USP10 in autophagy in NAFLD is unclear. Recently, positive feedback was found between USP10 and AMPK,20 and we speculated that USP10 could directly regulate autophagy to alleviate hepatic steatosis. In this study, HepG2 cells were treated with PA to model NAFLD in vitro. HepG2 cells are suitable experimental NAFLD models because their biological and genetic characteristics.21,22 PA is the most abundant unsaturated fatty acid in humans. USP10 expression was regulated by lentivirus infection. The results confirmed the underlying mechanism of the protective effects of USP10 on autophagic activity, autophagic flux, lipid-targeted autophagy in hepatic steatosis.
PA solution was purchased from KunChuang (cat. no. 4; Xi’an, China). Hematoxylin (cat. no. 517-28-2), Oil Red O (cat. no. G1260), Bafilomycin A1 (cat. no. A8510), and Earle’s balanced salt solution (EBSS) (cat. no. H2020) were purchased from Solarbio (Beijing, China). Chloroquine diphosphate salt (CQ) (cat. no. C6628) was purchased from Sigma-Aldrich (St. Louis, MO, USA). 3-Methyladenine (3-MA) (cat. no. HY-19312), rapamycin (cat. no. HY-10219), and DB07268 (cat. no. HY-15737) were purchased from MedChemExpress (Monmouth Junction, NJ, USA). Dulbecco’s modified Eagle’s medium (DMEM) (cat. no. 11965092), fetal bovine serum (cat. no. 12483020), and penicillin–streptomycin (cat. no. 15070063) were purchased from Gibco (Waltham, MA, USA). Phosphate buffered saline (PBS) (cat. no. G4202) was purchased from Servicebio (Wuhan, China).
HepG2 human hepatoma cells (GeneChem, Shanghai, China) were cultured in DMEM supplemented with 10% fetal bovine serum, 100 U/mL penicillin and 100 µg/mL streptomycin 37°C in humidified air containing 5% CO2. PA and vehicle were diluted in culture medium to 125–500 µM for treatment. Before PA treatment, HepG2 cells were incubated with CQ (50 µM for 6 h), 3-MA (5 mM for 6 h), EBSS (2 h), bafilomycin A1 (100 nM for 5 h) and rapamycin (1 µM for 5 h). During PA treatment, HepG2 cells were co-incubated with DB07268 (1–216 µM).
HepG2 cells were seeded in six-well plates at 2.5×105 cells/mL and treated with PA (125–500 µM) and vehicle for 24 h when they reached 50% confluence. Before PA or vehicle treatment, cells were pretreated with 3-MA or shATG5. After treatment, the cultures were washed three times with PBS, and the cells were fixed with 10% formalin for 20 m, washed three times with PBS, and stained with Oil Red O solution for 30 m. Cells were then washed three times with PBS and photographed. To quantify the extent of steatosis, LDs in cells were dissolved in isopropanol and absorbance of eluates at 520 nm was read with a microplate reader.
The concentration of triglycerides in HepG2 cells was determined with a triglyceride quantification kit (cat. no. E1013; Applygen, Beijing, China).
The cytotoxicity of PA, vehicle, and DB07268 was measured by cell counting kit-8 (CCK-8) assays. Briefly, 1×104 cells/well were seeded in 96-well plates overnight and treated by PA (125–500 µM), vehicle, or co-incubated with DB07268 (1–216 µM) for 24 h. CCK-8 (cat. no. C0037; Beyotime, Beijing, China) was added to each well and the plates were incubated for an additional 1 h. The absorbance of the cultures was measured at 450 nm with a microplate reader.
HepG2 cells were seeded in 12-well plates at 5×105 cells/mL and treated with PA (312.5 µM), vehicle, or co-incubated with DB07268 (9–45 µM) when they reached 70% confluence. Total RNA was extracted using RNAiso Plus (cat. no. 9108; Takara, Tokyo, Japan) PrimeScript RT reagent kit (Perfect Real Time, cat. no. RR037; Takara) was used to reverse transcribe RNA to cDNA. qPCR was performed on an Applied Biosystems 7500 Real-Time PCR System (Hercules, CA, USA) using TB Green Premix ExTaq (Tli RNaseH Plus, cat. no. RR420; Takara). The qPCR total reaction volume was 20 µL, including TB Green Premix ExTaq 10 µL, forward primer 10 μM 0.4 mL, reverse primer 10 µM 0.4 µL, ROX Reference Dye II (50×) 0.4 µL, DNA templates 50 ng/µL and sterile water 6.8 µL. qPCR cycling included three stages: 1, 95°C for 30 s; 2, 40 cycles of 95°C for 5 s, 55°C for 30 s and 72°C for 30 s; 3, 95°C for 15 s, 60°C for 60 s and 95°C for 15 s. Primers were provided by Tianyi Huiyuan Biotech Co. Ltd. (Beijing, China). The sequences are shown in Supplementary Table 1.
HepG2 cells were seeded in 12-well plates at 5×105 cells/mL and treated with PA (312.5 µM), vehicle, or co-incubated with DB07268 (9–45 µM) when they reached 70% confluence. Total protein was extracted using 1× sodium dodecyl sulfate (SDS)-polyacrylamide denaturing protein loading buffer (cat. no. B1012; Applygen) supplemented with 1× phosphatase inhibitor mixture (cat. no. P1260; Applygen). As LC3B-2 is easily degraded when outside cells, western blotting was performed after extraction of total protein. Equal amounts of cellular proteins were resolved by 6–15% SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes (cat. no. IPVH00010; Merck Millipore, Burlington, MA, USA). Membranes were blocked with 5% bovine serum albumen (cat. no. G5002; Servicebio, Wuhan, China) for 1 h or QuickBlock (cat. no. P0252; Beyotime) for 10 m and incubated overnight at 4°C with specific primary antibodies (1:1,000). Antibodies to glyceraldehyde-3-phosphate dehydrogenase (cat. no. GB11002) and α-tubulin (cat. no. GB11200) were provided by Servicebio. Antibody to β-actin (cat. no. AF5003) was provided by Beyotime. Antibody to USP10 (cat. no. ab109219) was provided by Abcam (Cambridge, MA, USA). Antibodies to LC3B (cat. no. 2775), lysosomal-associated membrane protein (LAMP)1 (cat. no. 9091), ATG14 (cat. no. 5504), mammalian target of rapamycin (mTOR) (cat. no. 2972) and p-mTOR at Ser2448 (cat. no. 2971) were provided by Cell Signaling Technology (CST) (Danvers, MA, USA). Antibodies to ATG5 (cat. no. A0203), Beclin1 (cat. no. A11761), tuberous sclerosis complex (TSC)2 (cat. no. A19540), C-jun N-terminal protein kinase-1 (JNK1) (cat. no. A0288), phosphatidylinositol 3-kinase (VPS34) (cat. no. A12295), B-cell lymphoma 2 (Bcl2) (cat. no. A19693), p-Bcl2 at Ser70 (cat. no. AP0575), ribosomal protein S6 kinase (S6K) (cat. no. A4898), p-S6K at Thr389 (cat. no. AP1059) and p62 (cat. no. A7758) were provided by ABclonal (Woburn, MA, USA). After incubation, membranes were probed with secondary antibodies (cat. no. A0208; Beyotime) for 1 h at room temperature. Signals were read with a super enhanced chemiluminescence detection kit (cat. no. NE0102; NovovBio, Beijing, China). Specific protein bands were visualized with a Syngene imaging system and gray values were measured with Image J software.
Lentiviral vectors were synthesized by Shanghai Jikai Gene Chemical Technology Co., Ltd (Shanghai, China). For USP10 RNA interference, three pairs of short hairpins were designed and their target sequences (5′ to 3′) were shUSP10 4-1 ccCATGATAGACAGCTTTGTT, shUSP10 5-1, ccTATGTGGAAACTAAGTATT, and shUSP10 6-1 gcTGTGGATAAACTACCTGAT. They were cloned into the lentiviral vector GV344. Interfering RNAs for ATG5 (effective target sequence 5′ to 3′ ccTGAACAGAATCATCCTTAA) and TSC2 (effective target sequence 5′ to 3′ cgACGAGTCAAACAAGCCAAT) were also synthesized. They were cloned into the lentiviral vector GV248. A recombinant lentiviral vector expressing a scrambled shRNA was used as a negative control. Lentiviral vectors caring USP10 sequence were also synthesized. Lentivirus particles were used to infect HepG2 cells and establish stable cell lines. To select positive cells, 2 µg/mL puromycin or flow-cytometric sorting was applied. Infection efficiency was confirmed by western blotting and Q-PCR.
HepG2 cells were seeded on in six-well plates at 2.5×105 cells/mL and treated with PA (312.5 µM) or vehicle for 24 h, or additional by EBSS, rapamycin, CQ, 3-MA, bafilomycin A1, or DB07260 when they reached 30% confluence. After treatment, HepG2 cells were harvested and washed three times with PBS. The cells were fixed with 4% paraformaldehyde for 20 m, washed with PBS, and stained with BODIPY 493/503 (cat. no. GC42959; GLPBIO, Montclair, CA, USA) 0.5 µg/mL at room temperature. Tyramide signal amplification (TSA) was used to detect immunofluorescence in cells. HepG2 cells were deparaffinized and rehydrated. After antigen retrieval, blocking of endogenous peroxidases, and blocking with serum, they were incubated primary antibodies: anti-LC3B 1:200 (cat. no. 3868; CST); anti-LAMP1 1:50 (cat. no. 9001; CST); anti-USP10 1:50 (cat. no. ab109219; Abcam); and anti-TSC2 1:50 (cat. no. A19540; ABclonal). They were then incubated with horseradish peroxidase-labeled secondary antibodies, Cy3-TSA, or fluorescein isothiocyanate conjugated-TSA, with microwave treatment, the second round of primary antibodies and secondary antibodies incubation, and spontaneous fluorescence quenching. After stained with Bodipy 493/503 or antibodies, the nuclei of HepG2 were stained with 4′,6-diamidino-2-phenylindole (DAPI) at room temperature for 10 m. The cells were washed three times with PBS, cover slipped using anti-fade mounting medium, and observed by fluorescence microscopy and photographed. Details of fluorescent staining are described in the Servicebio protocol.
The results were reported as means±SD and the sample size reported for each procedure corresponds to number of results obtained from at least replicates. Between-group comparisons were performed with independent sample t-tests. Comparisons among three or more groups were performed with one-way analysis of variance followed by Tukey’s post hoc test. Differences were considered statistically significant when p was <0.05. GraphPad Prism 8 software was used for the statistical analysis.
HepG2 cells were treated with PA (125–500 µM) for 24 h, and compared with vehicle treatment, the number of LDs (Oil red O staining) and the triglyceride concentration (triglyceride assay) in HepG2 cells were increased by PA (Supplementary Fig. 1A, B). PA induced cellular steatosis in HepG2 cells. LDs induce lipotoxic effects in cells, with release of reactive oxygen species (ROS), which promotes NAFLD progression as a precursor.23,24 A previous study found that ROS reaction was inducible when hepatocellular viability decreased to 50% under PA stimulation.25 In this study, cellular viability of HepG2 cells decreased in PA groups with dose-independence (Supplementary Fig. 1C). Compared with the blank control group, cell viability as reduced by 250–375 µM PA (Supplementary Fig. 1C). We thus used the mean of 312.5 µM as the PA concentration for the modeling NAFLD in subsequent experiments.
Cells were harvested at 6, 12, and 24 h during PA stimulation. Representative molecules were detected to reflect different stages of autophagy, including induction of autophagy. They were Beclin1 and ATG14), synthesis of autophagosomes [ATG5, ATG7, and unc-51 like autophagy activating kinase (ULK1),4 autophagosome marker (LC3B),4 substrate of autophagy (p62),4 and degradation of autophagosomes (LAMP1). USP10 gene expression decreased successively during PA stimulation (Fig. 1A). LC3B, p62, LAMP1, ULK1, ATG7, and ATG14 gene expression increased at 6–12 h and then decreased at 24 h (Fig. 1A). USP10 protein expression decreased significantly at 24 h. ATG5, Beclin1 and LC3B-2 protein expression increased at 6 h and then fell at 24 h. p62 decreased at 6 h followed by restoration at 24 h (Fig. 1B, C). The results indicate that autophagic activity increased soon after PA stimulation began and then decreased with time. A potential association of USP10 and autophagic activity may exist. We investigated the role of autophagy in HepG2 cells treated with PA. The autophagy inhibitor bafilomycin A1 enhanced LD accumulation (Fig. 1D) and the autophagy agonist EBSS reduced LD accumulation (Fig. 1D). These results indicate that autophagy plays an essential role in hepatocellular steatosis.
Autophagic flux reflects the dynamic process of autophagy process, which refers to synthesis of autophagosomes and their degradation. Blocking either one impairs autophagic flux, and fails to scavenge LDs in cells. Because CQ effectively increases lysosomal pH, and inhibits degradation of autophagosomes,26 previous studies have it has been used to investigate autophagic flux in previous studies.27,28 Supplementary Figure 2 shows how synthesis and degradation of autophagosomes can be detected with CQ. The effect of PA on autophagic flux in HepG2 cells was determined, and the average relative protein expression of LC3B-2 in each group is shown in Figure 2. In the presence of CQ, LC3B-2 expression was lower under PA stimulation compared with vehicle, suggesting that PA inhibited synthesis of autophagosomes. The decrease in LC3B-2 expression in the absence of CQ compared with the expression in the presence of CQ under PA stimulation was smaller compared with vehicle, suggesting that PA inhibited autophagosome degradation. Therefore, PA inhibited simultaneously inhibited the synthesis and degradation of autophagosomes.
We established HepG2 cells with USP10 overexpression (USP10-OE) or USP10 knockdown (USP10-KD) through lentivirus infection. Two batches of HepG2 cells infected with USP10-sequence (LV-USP10) were established, Genechem-1 and Genechem-2. Western blotting showed that, compared with the control (LV-NC), USP10 level in Genechem-2 was higher than that in Genechem-1 (Supplementary Fig. 3A). Therefore, Genechem-2 was used in subsequent procedures in the USP10-OE model. HepG2 cells infected with three shRNAs targeting the USP10 gene (LV-shUSP10) were established, i.e. 4-1, 5-1, and 6-1. qPCR and western blotting showed that, compared with the control (LV-shScram), USP10 expression was the lowest in 5-1 (Supplementary Fig. 3B). Therefore, 5-1 was used subsequent procedures in the USP10-KD model. Gain- and loss-of-function experiments were performed to ascertain whether USP10 regulated autophagic activity in HepG2 cells. qPCR showed that USP10-OE increased the expression of LC3B, p62, LAMP1, ULK1, and ATG14, but not ATG7 compared with PA alone (Fig. 3A). Western blotting showed that USP10-OE increased the expression of LC3B-2, LAMP1, ATG5, and Beclin1, but not p62 compared with PA alone (Fig. 3B). qPCR showed that USP10-KD suppressed the expression of LC3B, LAMP1, ATG7, ATG14, and p62, but not ULK1 compared with PA alone (Supplementary Fig. 4A). Western blotting showed that USP10-KD suppressed protein expression of LAMP1 and Beclin1, and promoted protein expression of p62 in contrast to PA-alone stimulation. Changes in LC3B-2 and ATG5 were not significant (Supplementary Fig. 4B). The results show that USP10 played a significant role in regulating autophagic activity of HepG2 cells under PA stimulation. 3-MA, an upstream inhibitor of autophagy,26 was used to investigate whether USP10 regulated autophagic activity through the classic pathway. The increase in expression of VPS34, ATG5, Beclin1, and LC3B-2 induced by USP10 was inhibited by 3-MA (Fig. 4A). Also, as the efficacy and toxicity of 3-MA were debatable, shATG5 was used to block autophagosome formation. shATG5 significantly suppressed LC3B and USP10 was overexpressed (Fig. 4B). The results reveal that USP10 regulated autophagic activity through the classic autophagy pathway.
Western blotting and immunofluorescence were used to determine the effects of gain- and loss-of-function of USP10 on autophagic flux in HepG2 cells.
Supplementary Figure 2 illustrates how CQ was used to detect autosome synthesis and degradation. Average relative protein expression of LC3B-2 in each group is shown in Figure 5A and B. In the presence of CQ, the amount of LC3B-2 increased when USP10 was overexpressed compared with the expression in controls, whether or not the cells were treated with PA. The result suggests that USP10 promoted autophagosome synthesis. With PA treatment, difference in LC3B-2 expression in the absence of CQ compared with expression in the presence of CQ was greater in cells overexpressing USP10 compared with controls, suggesting that USP10 promoted of autophagosome degradation (Fig. 5A). In the presence of CQ, whether or PA was also present, LC3B-2 expression decreased when USP10 was knocked down compared with controls, suggesting that lack of USP10 inhibited autophagosome synthesis. In the presence of PA and USP10 knockdown, the difference in LC3B-2 expression in the absence of CQ was smaller than the expression when CQ was present, in contrast to controls. The result suggests that lack of USP10 inhibited autosomal degradation (Fig. 5B) and that USP10 simultaneously increased autophagosome synthesis and degradation.
In the presence of CQ synthesized autophagosomes appear as red dots (LC3B) and lysosomes appear as green dots (LAMP1).29 Autolysosomes appear as yellow dots, i.e. degraded autophagosomes. USP10-OE increased the numbers of red and yellow dots (Fig. 5C, D), and USP10 knockdown resulted in decreases in the numbers of red and yellow dots decreased (Fig. 5C, D). USP10 thus increased autophagosome synthesis and degradation. Taken together, the results of western blotting and immunofluorescence confirmed the positive effect of USP10 on autophagic flux in HepG2 cells.
USP10-OE alleviated, and USP10-KD aggravated HepG2 cells steatosis (Fig. 6A), but the role of autophagy induced by USP10 in regulating intracellular LDs was not clear. The autophagy inhibitors bafilomycin A1, CQ, 3-MA, and shATG5 and agonists EBSS and rapamycin significantly reversed the effects of USP10-OE and USP10-KD on intracellular LDs (Fig. 6A, B). These results show that USP10 alleviated cellular steatosis that depended on autophagy. Specifically, USP10 regulated lipid-targeted autophagy. USP10-OE enhanced and USP10-KD abolished, colocalization of LC3B puncta and LAMP1 with LDs (Fig. 6C, D); USP10 thus promoted lipid-targeted autophagy in HepG2 cells.
We investigated whether USP10 regulated induction-stage autophagy. JNK1-Beclin1 and TSC2-mTOR are classic signaling pathways for inducing autophagy.30 In contrast to treatment with PA alone, USP10-OE increased TSC2, JNK1, VPS34, and ATG14 protein expression accompanied by upregulation of the p-Bcl2/Bcl2 ratio and downregulation of the p-mTOR/mTOR and p-S6K/S6K ratios (Fig. 7A, B). The results show that USP10 activated the JNK1-Becin1 and TSC2-mTOR signaling pathways in HepG2 cells. We also investigated whether USP10 induced autophagy that depended on JNK1. HepG2 cells were treated with DB07268, a specific inhibitor of JNK131 in the presence of PA. The concentration–viability curve is shown in Supplementary Figure 5, and the half maximal-inhibitory concentration (IC50) of 90.51 µM. Subsequently, DB07268 at 1/10 (9 µM), 1/5 (18 µM), and 1/2 (45 µM) the IC50 was used to treat HepG2 cells. When USP10 was overexpressed, DB07268 significantly decreased JNK1 with dose-dependence (Fig. 8), but the changes in LC3B-2 and TSC2 were not consistent with those observed for JNK1. DB07268 (9–18 µM) did not significantly change TSC2 expression, but did increase LC3B-2. DB07268 (45 µM) inhibited both TSC2 and LC3B-2, and DB07268 (9–18 µM) treatment significantly increased the TSC2/JNK1 ratio (Fig. 8). We speculate that the rebound increase in LC3B-2 during DB07268 (9–18 µM) treatment was driven by compensatoryTSC2 expression against lack of JNK1. The compensatory expression of TSC2 was not observed when JNK1 was entirely inhibited entirely, leading to a rapid decrease in LC3B-2. DB07268 treatment on HepG2 cells resulted in similar outcomes when USP10 was knocked down (Supplementary Fig. 6). The TSC2/JNK1 ratio was lower with DB07268 (9–18 µM) treatment than it was in cells overexpressing USP10 (Supplementary Fig. 6), which indicated that the degree of TSC2 compensatory expression was associated with the USP10 level. Immunofluorescence showed the strong colocalization of USP10 and TSC2 in HepG2 cells treated with DB07268 (18 µM) treatment (Supplementary Fig. 7). We thus speculate that the compensatory expression of TSC2 was directly mediated by USP10. We determined whether the TSC2-to-JNK1 compensation was required for the rebound increase in LC3B-2. When TSC2 was knocked down, LC3B-2 decreased rapidly with DB07268 treatment without the rebound increase (Supplementary Fig. 8A). TSC2-to-JNK1 compensation was thus required to maintain LC3B-2 expression. Of note, in the absence of DB07268, expression of JNK1 and LC3B-2 induced by USP10 were not affected by TSC2 knockdown (Supplementary Fig. 8B). The results show that JNK1 was involved in the autophagy induced by USP10, and that TSC2 had an important supplementary role in maintaining autophagy. We also investigated whether the TSC2-to-JNK1 compensation impacted the lipid-lowering effect of USP10. When TSC2 was knocked down, DB07268 abolished the lipid-lowering effect of USP10 in a dose-dependent manner (Supplementary Fig. 9). The results show that the JNK1/TSC2 signaling pathway was required for USP10 to induce autophagy and scavenge LDs in HepG2 cells.
NAFLD places an enormous burden on global healthcare. It can coexist with other metabolic syndromes including type 2 diabetes mellitus, obesity, and chronic cardiovascular disease, and contributes to a vicious cycle that increases liver-specific and overall mortality.1 Although the two-hit theory is a cornerstone for pathogenesis of NAFLD, details need to be addressed. USP10 is a novel mediator of NAFLD that inhibits hepatic steatosis, insulin resistance, and inflammation. In this study, USP10 had positive effects on autophagy in HepG2 cells that were stimulated by PA, including restoration of autophagic activity, promoting autophagic flux, increasing lipid-targeted autophagy, and activating signaling pathways that promoted autophagy. Of note, we discovered TSC2-to-JNK1 compensation for maintaining autophagy that was associated with USP10 level. The collective effects on autophagy are important pathway of USP10 to alleviate steatosis in hepatocytes. Autophagy has an essential role in NAFLD. Immunohistochemical staining of liver tissue from NASH patients has shown that the number of LC3B puncta was significantly decreased compared with normal controls.32 In a mouse NASH model induced by a methionine- and choline-deficient diet, transmission electron microscopy showed a decrease in the number of autophagosomes and LC3B-2 expression in liver tissue.33 In our study, HepG2 cells were used to model hepatocellular steatosis. PA suppresses autophagy, and leads to accumulation of LDs and lipotoxicity,34 and in our study, PA had time-dependent effects on autophagy in HepG2 cells. PA promoted autophagy at an early stage but autophagy was impaired later on. Similar results have been obtained in HL-7702, cells, L02 cells, and AML-12 cells.8,35 Adenovirus-mCherry-GFP-LC3B has been used to measure autophagic flux,36,37 but has some limitations. First, adenovirus-mCherry-GFP-LC3B cannot completely replace the endogenous LC3B in cells. Second, it is suitable for detecting autophagosome degradation but not synthesis. Co-immunofluorescence staining of LC3B puncta and LAMP1, and quantitative analysis of LC3B-2 protein expression,38 are novel approaches to discriminate between disturbed autophagic initiation and lysosome function that were used in this study to evaluate autophagic flux. PA inhibited synthesis and degradation of autophagosomes. Specifically, colocalization of LC3B puncta and LAMP1 with LDs showed that PA inhibited lipid-targeted autophagy. The result is consistent with a previous study in which DAPGreen and LysoTracker were used to detect the effect of oleic acid on lipid-targeted autophagy.10 Another study found that lipophagy was weakened in AML12 cells treated with oleic acid and PA, which supports our findings.35 USP10 plays is involved in tumorigenesis39–41 and energy metabolism16,20 in the liver. In this study, USP10 restored autophagy and promoted autophagic flux to alleviate hepatocellular steatosis. Of note, USP10 upregulated p62 gene expression, but did not affect its protein expression. We speculate that p62 was degraded during autophagic flux after its translation. In fact, autophagy cannot be evaluated based only on p62. p62 is a substrate protein and mediates binding of autophagosomes and their cargo.4 Impaired autophagy induces p62 accumulation,10,35 and a previous study found that p62 itself directly induced autophagy.42 Therefore, it is hard to distinguish p62 protein as an outcome of impaired autophagy or as a requirement for autophagy activation. USP10 regulates downstream molecules by post-translational modification, and a recent study reported that USP10 reduced the ubiquitination of LC3B-2 in H4 cells.19 Other explanations of the how USP10 may have increased LC3B-2 include elevated LC3B gene expression in response to USP10 overexpression. A second explanation is that ATG5 and ATG7 gene and protein expression, which are involved in synthesis and maturity of LC3B-2 protein, were regulated by USP10. Third, USP10 promoted the expression of LC3B-2 protein, with dependence on VPS34 and ATG5. Signaling pathways associated with inducing autophagy were identified in our study. In the mTOR signaling pathway, USP10 decreased the p-mTOR/mTOR and p-S6K/S6K ratios in HepG2 cells. A previous study found that USP10 functioned as a tumor suppressor in hepatocellular carcinoma cells, stabilizing AMPKα by inhibiting its polyubiquitylation and negative regulation of mTOR.43 USP10 has been found to increase Beclin1, VPS34 and ATG14 simultaneously. USP10-OE was reported to reduce the levels of ubiquitinated Beclin1 in H4 and 293T cells, without affecting the ubiquitination of VPS34, ATG14 and p150, were not affected.18 Post-translational modifications may thus be important, but not a unique mechanism, of the USP10 regulation of autophagy in HepG2 cells exposed to PA. JNK1 and TSC2 are molecules upstream of Beclin1 and mTOR, and a JNK1 inhibitor was used to determine whether USP10 induced autophagy required JNK1. Previous studies have used SP600125 to inhibit JNK1-induced autophagy,44,45 but SP600125 inhibits other JNK family members (JNK1, JNK2 and JNK3). DB07268 is a specific JNK1 inhibitor, and it revealed that USP10 induced autophagy that depended on JNK1 rather than TSC2. TSC2-to-JNK1 compensation was seen to maintain autophagy to some extent when JNK1 was inhibited. The extent of TSC2-to-JNK1 compensation was associated with the USP10 level. However, possible mechanisms of interaction among USP10, JNK1 and TSC2 have not been previously reported. In our study, autophagic inhibitors and shATG5 decreased the protective effect of USP10 on LD accumulation. Lipid-targeted autophagy was promoted when USP10 was overexpressed, which makes autophagy a novel pathway for USP10 to alleviate hepatocellular steatosis. Lack of JNK1/TSC2 caused severe steatosis in HepG2 cells even if USP10 was overexpressed. As a consequence, USP10 had lipid-lowering effects in HepG2 cells through JNK1/TSC2-induced autophagy.
USP10 alleviated PA-induced hepatocellular steatosis through autophagy, USP10 restored autophagy, promoted autophagic flux, and increased lipid-targeted autophagy. JNK1/TSC2 signaling pathways were required for USP10 to achieve its lipid-lowering effect. Some limitations exist. First, experimental outcomes were only observed in HepG2 cells. Second, research on lipotoxicity, inflammation, and fibrosis induced by autophagy was lacking. Third, other special types of autophagy such as lipophagy and mitophagy were not investigated. The effect of USP10 on autophagy should be investigated in NAFLD rodent models and patients. Additional mechanisms of the regulation of JNK1, TSC2, and ATGs by USP10 should be investigated.
Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. |
|
PMC9647105 | Yu-Fei Qin,Zi-Yu Zhou,Hou-Wei Fu,Hao-Ming Lin,Lei-Bo Xu,Wen-Rui Wu,Chao Liu,Xiao-Lin Xu,Rui Zhang | Hepatitis B Virus Surface Antigen Promotes Stemness of Hepatocellular Carcinoma through Regulating MicroRNA-203a | 26-07-2022 | Hepatitis B surface antigen,Hepatocellular carcinoma,microRNA,Stemness | Background and Aims Patients with persistent positive hepatitis B surface antigen (HBsAg), even with a low HBV-DNA load, have a higher risk of hepatocellular carcinoma (HCC) than those without HBV infection. Given that tumor stemness has a critical role in the occurrence and maintenance of neoplasms, this study aimed to explore whether HBsAg affects biological function and stemness of HCC by regulating microRNA, and to explore underlying mechanisms. Methods We screened out miR-203a, the most significant down-regulated microRNA in the microarray analysis of HBsAg-positive samples and focused on that miRNA in the ensuing study. In vitro and in vivo functional experiments were performed to assess its regulatory function. The effect of miR-203a on stemness and the possible correlation with BMI1 were analyzed in this study. Results MiR-203a was significantly down-regulated in HBsAg-positive HCC with the sharpest decrease shown in microarray analysis. The negative correlation between miR-203a and HBsAg expression was confirmed by quantitative real-time PCR after stimulation or overexpression/knockdown of HBsAg in cells. We demonstrated the function of miR-203a in inhibiting HCC cell proliferation, migration, clonogenic capacity, and tumor development in vivo. Furthermore, the overexpression of miR-203a remarkably increases the sensitivity of tumor cells to 5-FU treatment and decreases the proportion of HCC cells with stem markers. In concordance with our study, the survival analysis of both The Cancer Genome Atlas database and samples in our center indicated a worse prognosis in patients with low level of miR-203a. We also found that BMI1, a gene maintains the self-renewal capacity of stem cells, showed a significant negative correlation with miR-203a in HCC specimen (p<0.001). Similarly, opposite BMI1 changes after overexpression/knockdown of miR-203a were also confirmed in vitro. Dual luciferase reporting assay suggested that miR-203a may regulate BMI1 expression by direct binding. Conclusions HBsAg may promote the development of HCC and tumor stemness by inhibiting miR-203a, resulting in poor prognosis. miR-203a may serve as a crucial treatment target in HBsAg-positive HCC. More explicit mechanistic studies and animal experiments need to be conducted as a next step. | Hepatitis B Virus Surface Antigen Promotes Stemness of Hepatocellular Carcinoma through Regulating MicroRNA-203a
Patients with persistent positive hepatitis B surface antigen (HBsAg), even with a low HBV-DNA load, have a higher risk of hepatocellular carcinoma (HCC) than those without HBV infection. Given that tumor stemness has a critical role in the occurrence and maintenance of neoplasms, this study aimed to explore whether HBsAg affects biological function and stemness of HCC by regulating microRNA, and to explore underlying mechanisms.
We screened out miR-203a, the most significant down-regulated microRNA in the microarray analysis of HBsAg-positive samples and focused on that miRNA in the ensuing study. In vitro and in vivo functional experiments were performed to assess its regulatory function. The effect of miR-203a on stemness and the possible correlation with BMI1 were analyzed in this study.
MiR-203a was significantly down-regulated in HBsAg-positive HCC with the sharpest decrease shown in microarray analysis. The negative correlation between miR-203a and HBsAg expression was confirmed by quantitative real-time PCR after stimulation or overexpression/knockdown of HBsAg in cells. We demonstrated the function of miR-203a in inhibiting HCC cell proliferation, migration, clonogenic capacity, and tumor development in vivo. Furthermore, the overexpression of miR-203a remarkably increases the sensitivity of tumor cells to 5-FU treatment and decreases the proportion of HCC cells with stem markers. In concordance with our study, the survival analysis of both The Cancer Genome Atlas database and samples in our center indicated a worse prognosis in patients with low level of miR-203a. We also found that BMI1, a gene maintains the self-renewal capacity of stem cells, showed a significant negative correlation with miR-203a in HCC specimen (p<0.001). Similarly, opposite BMI1 changes after overexpression/knockdown of miR-203a were also confirmed in vitro. Dual luciferase reporting assay suggested that miR-203a may regulate BMI1 expression by direct binding.
HBsAg may promote the development of HCC and tumor stemness by inhibiting miR-203a, resulting in poor prognosis. miR-203a may serve as a crucial treatment target in HBsAg-positive HCC. More explicit mechanistic studies and animal experiments need to be conducted as a next step.
Hepatocellular carcinoma (HCC) is the most common fatal primary liver cancer.1 As in many other solid tumors, cancer stem cells (CSCs) have an essential role in the progression of HCC.2,3 According to CSC hypothesis, a minority cell population in cancer with the property of extensive self-renewal contributes to tumor growth and heterogeneity. With regard to HCC, CSCs are identified by canonical cell surface markers including CD133, CD90, CD44, BMI1, EpCAM and oval cell marker OV6, etc.4–6 HBV infection is a critical risk factor for liver cirrhosis and HCC,7 as it accounts for about 80% of all HCC cases worldwide and increases the risk of HCC approximately by 20 times.8,9 Although HBV infections are effectively controlled by long-term nucleoside analog therapy, they are a public health problem in HCC carcinogenesis, with risk paralleled by the HBV virus burden.10 It has been reported that patients with residual hepatitis B virus surface antigen (HBsAg) titers higher than 1,000 IU/mL are much more likely to suffer from HCC.11 However, the exact influence of HBsAg on HCC is not well documented. Chisari et al.12 reported a greater probability of HCC in HBsAg transgenic mice. Recently, we found that the stem cell marker, oncogene B cell-specific Moloney murine leukemia virus integration site 1 (BMI1) was overexpressed in HBsAg transgenic mouse liver and HBsAg-positive human HCC tissue.13 We suspected that BMI1 be an underlying mechanism in HBsAg-induced HCC, as BMI1 is an important cofactor of polycomb repressive complex 1 associated with cell cycle regulation, cell apoptosis, and maintenance of stem cell self-renewal.14 Increased expression of BMI1 was observed to maintain the tumor-initiating ability of human HCC,15–18 consistent with our findings in HCC with bile duct tumor thrombi.19 We also demonstrated that BMI1 knockdown decreased proliferation, colony formation, and invasiveness of human HCC cells in vitro and significantly increased its chemosensitivity.20,21 Most importantly, we showed that forced expression of BMI1 promoted the malignant transformation of rat liver progenitor cells into liver CSCs.22,23 The findings indicated the importance of BMI1 in the maintenance of stemness and the potential role in initiating cancer. Micro RNAs (miRNAs) are short noncoding RNAs with 18–22 nucleotides that regulate gene expression by interfering with endogenous RNA machinery.24 Cumulative evidence has demonstrated that miRNAs can function as tumor promoters or suppressors, and regulate biological processes including apoptosis, invasion, proliferation, and stemness.25–27 This study investigated the biological function of miRNAs in HCC with high HBsAg titers.
A cohort of 55 paired frozen liver tumor and normal adjacent tissues four fresh tumor tissues were obtained from HCC patients undergoing surgical resection at Sun Yat-sen Memorial Hospital. Patients who had received radiation therapy or chemotherapy prior to radical tumor resection were excluded. The difference and significance of HBsAg, miR-203a, and BMI1 expression in these patients were investigated. Routine histological examination of hematoxylin and eosin stained tissue confirm the HCC diagnosis.
Four fresh HCC tissues with or without HBV were obtained from HCC patients undergoing surgical resection at Sun Yat-sen Memorial Hospital. Total RNA was extracted with TRIzol reagent (TaKaRa, Shiga, Japan). RNA was purified with mirVana miRNA isolation kits (Ambion, Austin, TX, USA), tailed with polyadenosine-acidified polymerase, combined with biotinized DNA dendritic polymers, and then hybridized to Affymetrix GeneChip miRNA arrays using FlashTag Biotin RNA labeling kits (Genisphere, Hatfield, PA, USA). We used an Affymetrix GeneChip Scanner 3000 (Affymetrix, Santa Clara, CA, USA) to scan slides and the miRNA QC Tool to analyze miRNA data.
Huh-7, HepG2, HepG2.2.15, HepG2.117 HCC cell lines stably transfected with HBV genome13,28 and LO2 human liver cells were obtained from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). All cells were cultured in Dulbecco’s Modified Eagle Medium (Gibco, Shanghai, China) supplemented with 10% fetal bovine serum (Biological Industries, Beit HaEmek, Isreal) and 1% penicillin-streptomycin (New Cell & Molecular Biotech, Suzhou, China) at 37°C in a humidified incubator with 5% CO2. G418 was needed for stably screening of HepG2.2.15 and hygromycin was used for HepG2.117 culture.
We collected HBV-containing supernatants with 6% polyethylene glycol 8000 (Sigma, Darmstadt, Germany) precipitated at 4°C overnight, and concentrated by centrifugation at 12,000 g for 60 m at 4°C. The supernatants of HepG2 cells collected by the same procedure served as the non-particle control. HCC or LO2 cells were seeded in 24-well plates and incubated with HBV particles (MOI>1,000) for 24, 48, 72 h, and 4 days. At the end of the incubation, cells were further cultured in normal maintenance medium.
We stimulated Huh7 and HepG2 cells 500 ng/mL and LO2 cells with 250 ng/mL recombinant HBsAg adr (HBS-875, PROSPEC, Israel) for 5 days, with daily change of the medium and addition of HBsAg adr to maintain the concentration. The procedures followed the supplier’s instructions, using isovolumetric phosphate buffered saline (PBS) as the control. RNA was collected 5 days after stimulation for further use.
HBsAg-encoded plasmid pCDNA-HBsAg and pCDNA3.1 vector, which were provided by Professor Mengji Lu, were transiently transfected into three cell lines (HepG2, Huh-7, and LO2).29 SiRNA-HBsAg was obtained from GenePharma (Shanghai, China) and transiently transfected into HepG2.2.15 HCC cells transfected with the total HBV genome. Hsa-miR-203a-mimics (5′-GUGAAAUGUUUAGGACCACUAG-3′); hsa-miR-203a-inhibitors (5′-CUAGUGGUCCUAAACAUUUCAC-3′), were transiently transfected into two HCC cell lines (HepG2 and Huh-7) using Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) following the manufacturers’ recommendations.22 The cells were harvested 48 h after transfection for subsequent procedures.
Total RNA was extracted with RNAiso Plus reagent (TaKaRa). The primers are shown in Supplementary Table 1. Glyceraldehyde phosphate dehydrogenase (GAPDH) was used as the housekeeping gene and relative expression was calculated with the 2−ΔΔ cycle threshold methods. qPCR was performed following the manufacturer’s recommendations as previously described.23
Cell pellets were lysed with RIPA buffer (Beyotime, Beijing, China). Protein concentration was determined using a bicinchoninic assay kit (Beyotime). Samples were denatured in 5× loading buffer at 95°C for 10 m. Proteins were separated by 10% sodium dodecyl-sulfate polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes. The membranes were blocked with 5% dried skimmed milk in tris-buffered saline with Tween (TBST) at room temperature for 1 h and incubated with primary antibodies recognizing BMI1 (1:300) and GAPDH (1:1,000) overnight at 4°C. The membranes were incubated with the appropriate HRP-conjugated secondary antibodies (1:5,000) next day after washed by TBST three times. The proteins were visualized with by enhanced chemiluminescence plus reagents (Beyotime). The primary and secondary antibodies are listed in Supplementary Table 2.
HCC cells were seeded into 96-well plates at the density of 2×103/well and four repetitive wells for each group. After 6–8 h the cells were treated with 10% cell counting kit-8 (CCK-8; Dojindo Molecular Technologies) in an incubator for 1 h. The absorbance at 450 nm was measured with an ELISA plate reader and cell growth (%) was calculated.
Suspensions of 1,000 HCC cells were seeded into six-well plate and incubated in complete medium, which was changed every 3 days. After 14 days, the cells were fixed with 10% formaldehyde and stained with 0.1% crystal violet for colony counting.
An aliquot of 1×105 cells in 0.1 mL serum-free medium was placed in the upper chamber of 6.5 mm, 8 µm pore size polycarbonate membrane that was precoated with extracellular matrix gel (Corning, NY, USA). The lower chamber was loaded with 0.5 mL of medium containing 20% fetal bovine serum. The cells were fixed with 10% paraformaldehyde after 48 h incubation and then counterstained with 0.1% crystal violet. Cells migrating to the lower chamber were observed by light microscopy, and the number of migrating cells was calculated.
HCC cells were resuspended in PBS at a density of 1×107 cells/100 µL and stained with annexin V-FITC, propidium iodide-phycoerythrin (Abcam, Cambridge, UK), anti-human CD133-PE antibody and CD90-PE antibody (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany), followed by incubation for 20 m on ice. The respective isotype controls were set up at the same concentrations. The cells were analyzed on by flow cytometry (FACSVerse; BD Biosciences, Franklin Lakes, NJ, USA).
ALDEFLUOR kits (Stem Cell Technologies, Vancouver, Canada) were used following the manufacturer’s instructions. HCC cells were suspended in Aldefluor assay buffer containing ALDH substrate (BAAA, BODIPY amino acetaldehyde, 1 mmol/L) at 1×106 cells/mL for 30 m, with or without the specific ALDH inhibitor diethylamino benzaldehyde (1 mmol/L). DEAB was used as an internal negative control for each individual experiment to distinguish between high ALDH activity (ALDH positive) cells and cells with low ALDH activity (ALDH negative). Analysis and sorting were conducted fluorescence-activated cell sorting. Aldefluor was excited at 488 nm and fluorescence emission was detected at 530/30. The data were analyzed by Cell Quest Pro and FlowJo (Ashland, KY, USA).
The HCC cell lines HepG2 and Huh7 cells were infected with lentiviruses and negative control purchased from GenePharma. Virus-containing medium was replaced with fresh culture medium after 12–24 h of infection. Virus-containing medium was replaced with fresh culture medium after 12–24 h of infection and then replaced again with fresh media containing puromycin after 48–96 h of infection to select for stably transfected cells. BMI1 mRNA and protein expression were verified by qPCR and western blotting.
All animal procedures were performed following protocols approved by the Institutional Animal Care and Use Committee at Sun Yat-sen University and were performed at Sun Yat-sen University (Guangzhou, China). Briefly, 5×106 HCC cells overexpressing miR-203a and control cells were suspended in 100 µL PBS-Matrigel (1:1) and injected subcutaneously into 3-4-week-old mice (Balb/c nu/nu), beginning 7 days after injection, tumor volumes were calculated every second day after measuring the length and width with calipers. Mice were sacrificed 5 weeks after injection, and the tumors were removed and measured. Volume in cm3 was calculated as (width2)×length/2.
The region of the human BMI13′-untranslated region (UTR; bases 8,334–10,276) contained three putative miR-203a-binding sites that were predicted by miRDB, miRmap, and TargetScan. A wildtype (WT) and four mutant (mut) 3′-UTR fragments of human BMI1 mRNA were amplified and subcloned to XhoI and NotI restrictive sites in the psiCHECK-2 vector (Applied Biosystems, Foster City, CA, USA), downstream of the fluorescent enzyme reporter gene. The primer sequences of BMI1 3′-UTR amplification are listed in Supplementary Table 3 (BMI1-3′-UTR-mut4, all three putative miR-203a-binding sites were mutated by double mutation). All clones were sequenced to verify the correctness of the nucleotide sequences. Luciferase activity was assayed with a dual luciferase reporter assay system (Promega, Madison, WI, USA).
The statistical analysis was performed with SPSS 17.0 (SPSS, Chicago, IL, USA). In vitro data were analyzed with Student’s t-tests; Mann-Whitney U-tests and log-rank tests were used to analyze in vivo data and clinical parameters. P-values <0.05 were considered statistically significant. All cell culture experiments were repeated at least three times independently, with three multiple wells at a time. The results were reported as means±SD.
Genome-wide miRNA expression profiles of HBV-positive and HBV-negative HCC tissues were utilized to evaluate HBV-related miRNAs. Among the expression patterns, we found miR-203a was significantly decreased in HBV-positive HCC tissues (Fig. 1). Further validation using qPCR confirmed the alteration of miR-203a in tissue specimens from 55 HCC patients. As shown in Figure 2A, miR-203a expression was significantly lower in HBV-positive than in HBV-negative tissues (p<0.01). We also found a negative correlation between HBsAg level and miR-203a expression in HBV-infected HCC tissues, which prompted exploration the underlying regulatory mechanism and the potential clinical transformation (Fig. 2B, p=0.029).
qPCR confirmed that HBV inhibited the of miR-203a expression in LO2 and Huh7 and HepG2 HCC cells (Fig. 2C–E). Stimulation by exogenous HBsAg protein, and overexpression, and knockdown experiments showed that HBV down-regulated miR-203a through HBsAg in HCC. Five days after stimulation with recombinant HBsAg, qPCR revealed decreased miR-203a expression in LO2 (p=0.0025), Huh7 (p=0.0043), and HepG2 cells (p<0.0001, Fig. 2F–H). HBsAg-coding plasmid was transfected into LO2 liver cells and HepG2 and Huh-7 HCC cells. siRNA-HBsAg was transfected into HepG2.2.15 HBV-related HCC cells. qPCR results showed that miR-203a was down-regulated after transfection with HBsAg plasmid forced overexpression in both HCC cell lines and in liver cells (Fig. 2F). miR-203a was upregulated in HepG2.2.15 cells by transcription with HBsAg-siRNA, compared with the control (Fig. 2G). The findings indicate that miR-203a expression in HCC cells was decreased by HBsAg in vitro (p<0.01).
Inspired by the significant down-regulation of miR-203a expression in HCC tissues and HCC cells and the clinical correlation, we investigated the biological significance of miR-203a in hepatocarcinogenesis. Control lentiviral vector-infected cells and miR-203a mimic vector-infected cells were injected subcutaneously into nude mice in opposite flanks. The tumor volumes were significantly smaller in in nude mice injected with cells overexpressing miR-203a (Fig. 3A, B, p<0.01). CCK-8 cell proliferation assays revealed that miR-203a mimics significantly decreased the proliferation of HCC cells (Fig. 3C, D). As shown in Figure 3E, F, miR-203a mimic-transfected HCC cells formed fewer and smaller colonies compared with the control group. Similarly, the invasiveness of miR-203a overexpressing cells was significantly attenuated (Fig. 3G, H). Overall, the results showed that endogenous overexpression of miR-203a inhibited the tumorigenicity of HCC cells both in vitro and in vivo, implying that miR-203a suppressed the self-renewal and invasiveness of HCC cells.
To assess the stem cell properties in HCC, classical markers CD133, and CD90 were evaluated by flow cytometry. The results revealed that forced expression of miR-203a decreased the percentage of CD133-positive Huh7 cells (28.34%±1.82% vs. 91.17%±3.83%, p<0.01). Representative images are shown in Figure 4A. However, there was no significant change in CD90 positivity in Huh7 cells that overexpressed miR-203a (3.53%±0.32 vs. 2.07%±0.33%, Fig. 4B). The ALDH assay results found that the average percentage of ALDH positive malignant stem cells was 63.02±6.29% in Huh7 cells overexpressing miR-203a and 85.74±1.46% in control cells (p<0.01). In HepG2 cells the 64.44±9.28% of the miR-203a upregulated cells and 83.49±1.69% of the control cells were ALDH positive (P=0.0249). Representative fluorescence-activated cell sorting images are shown in Figure 4C–F. Overall, the findings provide evidence of the regulation of hepatic CSC-like phenotypes of HCC cells by miR-203a.
To determine whether miR-203a up-regulation increased the sensitivity of HCC cells to chemotherapy, we studied the effect of miR-203a on cell apoptosis after 5-FU administration. Compared with the controls, miR-203a overexpression reduced HCC cell viability after 5-FU treatment (Fig. 5A, B). There was a great increase in the apoptosis rate of HCC cells transfected with miR-203a mimics in response to 5-FU, from 23.41±2.03% and 37.48±2.97%. The corresponding rates in NC-transfected HCC control cells were 9.72±1.2% and 18.06±2.07%. Representative images are shown in Figure 5C, D. The findings indicate that miR-203a overexpression sensitized HCC cells to chemotherapy drug-induced apoptosis (p<0.01).
We previously reported that the expression of BMI1 was significantly upregulated in human HCC tissues and in four human HCC cell lines.13,22 In consideration of its role in stemness maintenance and the biological effect consistent with miR-203a, we asked whether BMI1 was involved in miR-203a regulation. qPCR results indicated a negative correlation between expression of BMI1 mRNA and miR-203a in human HCC tissues (Fig. 6A, r=−0.469, p=0.04). Analysis of Starbase data30 also found a significant negative correlation of BMI1 expression and miR-203a (Fig. 6A, r=−0.242, p<0.001).
To find the underlying effect of miR-203a on BMI1, we examined the change of BMI1 in cells with altered miR-203 expression. The results of western blot analysis suggested that miR-203a overexpression significantly decreased the level of BMI1 in HCC cells, and miR-203a knockdown increased BMI1 expression (Fig. 6B). The luciferase assay was used to determine whether BMI1 was a target gene of miR-203a. Cotransfection of miR-203a mimics decreased luciferase expression of the BMI1-3′-UTR-WT reporter but had no effect on luciferase activity in the four BMI1-3′-UTR mutant reporters and the psiCHECK-2 control reporter. Similarly, only the luciferase expression of the BMI1-3′-UTR-WT reporter was elevated after cotransfection of miR-203 inhibitor (Fig. 6C). The results indicate that miR-203a served as a direct negative regulator of BMI1 expression in HCC cells.
To explore the role of miR-203a in the development of human liver cancer, we assayed miR-203a levels in HCC tissues. miR-203a expression was significantly lower in human HCC tissues than in adjacent normal tissues (Supplementary Fig. 1E, n=29, p<0.01). Analysis of the correlation of miR-203a expression and clinical parameters in clinical cases found that loss of miR-203a expression was correlated with larger tumor size (Table 1, p=0.03). Kaplan-Meier analysis (Fig. 6D) found that miR-203a was an independent prognostic factor of the overall survival of HCC patients (p=0.0362), consistent with the results in the database (Fig. 6D, p=0.036).31
Despite advanced medical interventions, HBV remains a major threat to public health as it is difficult to completely eliminate, persists in the liver, and plays multiple roles in different stages of hepatitis, liver cirrhosis, and liver cancer. Recently, evidence have accumulated that microRNAs regulate tumor-associated genes and are an integral part in some pathways, implying critical involvement in the progression of cancer.32 To distinguish microRNAs essential for the development of HBV-HCC, we designed a sequencing protocol to compare the microRNAs differentially expressed in HBV-positive and HBV-negative HCC tumor tissues, excluding paratumoral tissues. Most candidate microRNAs selected from the microarray were down-regulated in HBV-positive HCC tissues, including miR-30b-5p, miR-98-5p, miR-148a-3p, and miR-221-5p, which have been shown to regulate the development of HCC.33–37 miR-203a, with the most significant difference, has been reported to inhibit the growth and metastasis of HCC by regulating molecules such as interleukin-24 and matrix metalloprotein-2, and also negatively regulates the epithelial-mesenchymal transition process caused by HCV core protein.38–41 However, the relationship between HBV and miR-203a has not yet been described, and most studies have focused on cell proliferation and cell cycle changes rather than on tumor stemness. Among the well-established proteins of HBV, HBsAg is often underestimated and is considered to possess only weak cancer promoting activity, but increasing evidence shows that HBsAg is an independent risk factor for HCC. The aim of this study was to explore whether miR-203a was regulated by HBsAg and affect the tumor stemness of HCC. First, we used clinical specimens to demonstrate that miR-203a was associated not only with HBV infection but was also negatively correlated with HBsAg titers. Furthermore, miR-203a expression was negatively regulated by HBsAg, as shown by the results of HBV particle stimulation, exogenous HBsAg protein stimulation, and HBsAg overexpression in cell lines without HBV infection and in HBsAg knockdown of cell lines with intact HBV genomes. In the next step, we validated the tumor suppressor effect of miR-203a by functional experiments in vitro and in vivo and explored its influence on HCC stemness. In the evaluation of tumor stemness, we preferred the traditional markers CD133 and CD90. CD133 is one of the earliest markers of human hepatic CSCs in HCC. Not only does it increase the incidence of HCC, but drives it to a tumor with poor differentiation and promotes cell proliferation by regulating cell cycle-related pathways. It has been found that tissues with viral hepatitis shared the same expression level of CD133 as HBV-HCC, which leaves us a clue that HBV infection might enhance the stemness of HCC through CD133. Furthermore, CD133 is proven to act widely on cancer stemness-related genes, including NANOG, the SOX family, and BMI1, which provides a basis for our research on the role of BMI in HBV-HCC.42 On the other hand, normal hepatic progenitor cells express CD90, and its abnormal elevation affects the interaction and adhesion of cells, further accelerates the development of HCC, and increases its tolerance to chemotherapy.43 It has also been reported that the PreS1 of HBV genome increases the CD133+/CD90+ ratio in HCC.44 These two markers are thus used to assess HCC stemness. As shown in Supplementary Figure 2, the RNA levels of stemness-related genes in LO2 were increased after recombinant HBsAg adr stimulation, and were elevated in Huh7 and HepG2, but the differences were not significant. The reason may be that 5-day stimulation was too short for the alteration of cancer stemness. Given that tumor stemness only existed in a small fraction of cells, which was possibly masked by PCR or western blotting. Thus, we chose stably transfected cell lines to evaluate the effect of miR-203a on cancer stemness and utilized a more accurate method, flow cytometry, to detect changes in the proportion of stem marker-positive cells. The stemness of tumors can affect multiple aspects of biological function, and the one most closely related to clinical treatment is chemosensitivity.45–47 The effect of miR-203a on chemosensitivity of HCC was preliminarily verified with a traditional chemotherapeutic agent 5-FU, which inspired us to further investigate the underlying mechanism. The human BMI1 gene is a core component of polycomb inhibition complex 1 (PRC1) and mediates gene silencing by monoubiquitination of histone H2A. It has been reported to promote tumorigenesis by regulating the cell cycle inhibitory genes, p16 and p19 and is regarded as a marker of tumor stemness.48–50 We previously showed that BMI1 expression was upregulated in HBsAg-positive HCC.13,22 In this subsequent study, we showed that miR-203a inhibited BMI1 expression and the direct binding sites of miR-203a and BMI1 by dual luciferase reporter assay. The data supported the hypothesis that miR-203a affected HCC stemness through BMI1, but further studies are required to draw definite conclusions. Finally, we emphasize the translational value of our research, as miR-203a might be a prognostic marker of HBs-HCC. Clinical specimens from our center and patients in The Cancer Genome Atlas (TCGA) database, both indicated shorter overall survival of patients with low miR-203a expression. Our analysis of clinical features also found larger tumor diameters in those with low level miR-203a expression. However, patients with high miR-203a expression presented higher alpha fetoprotein (AFP) levels, which implied that miR-203a may affect tumor differentiation and warrants further investigation. In conclusion, our findings indicate that HBsAg promotes stemness and chemoresistance in HCC by regulating the microRNA-203a/BMI1 axis. Our data provide a hint of the mechanism of HBsAg in promoting HCC, but more evidence is required for confirmation.
Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. |
|
PMC9647106 | Cyriac Abby Philips,Rizwan Ahamed,Jinsha K.P. Abduljaleel,Sasidharan Rajesh,Philip Augustine | Identification and Analysis of Gut Microbiota and Functional Metabolism in Decompensated Cirrhosis with Infection | 13-06-2022 | Microbiome,Portal hypertension,Sepsis,Acute-on-chronic liver failure,Multidrug resistance | Background and Aims Intestinal dysbiosis play a role in the adverse outcomes of sepsis and septic shock. However, variations in bacterial diversity and microbiota-related functional metabolic alterations within the gut microbiome in decompensated cirrhosis (DC) patients with infection remain unknown. Methods We conducted 16-srRNA sequencing on stool samples (n=51: sepsis, 27/no sepsis, 24) collected from consecutive DC patients upon admission. Bacterial diversity, significant taxa, and respective metabolic profiling were performed based on subgroup comparisons. Conet/Cytoscape was utilized to identify significant non-random patterns of bacterial copresence and mutual exclusion for clinical events. Results Genera associated with pathogenicity in conditions of immune exhaustion (Corynebacterium, Lautropia) were predominant in patients with sepsis. Metabolic pathways associated with oxidative stress and endotoxemia [lipopolysaccharide (LPS) synthesis and sulfur relay] were significantly upregulated in sepsis. Specific taxa were associated with sites of infection in DC patients. Protective oxidant pathways that increase glutathione were upregulated in those without sepsis. Gammaproteobacteria family of sulfur-metabolizing bacteria, exaggeration of orally predominant pathogens (Prevotella), and pathways of severe LPS-related hyperinflammatory stress were notable in those with interleukin-6 levels >1,000 pg/dL. Pathogenic genera related to an immune deficient state was significant in DC with ≥2 infection episodes. Megamonas was associated with survival during the same admission. Conclusions Specific gut microbiota and their metabolites were associated with sepsis and related events in patients with DC. Identifying beneficial strains that reduce immune exhaustion and supplementation of favorable metabolites could improve therapeutics for DC and sepsis, for which larger prospective, well controlled population-based studies remain an unmet need. | Identification and Analysis of Gut Microbiota and Functional Metabolism in Decompensated Cirrhosis with Infection
Intestinal dysbiosis play a role in the adverse outcomes of sepsis and septic shock. However, variations in bacterial diversity and microbiota-related functional metabolic alterations within the gut microbiome in decompensated cirrhosis (DC) patients with infection remain unknown.
We conducted 16-srRNA sequencing on stool samples (n=51: sepsis, 27/no sepsis, 24) collected from consecutive DC patients upon admission. Bacterial diversity, significant taxa, and respective metabolic profiling were performed based on subgroup comparisons. Conet/Cytoscape was utilized to identify significant non-random patterns of bacterial copresence and mutual exclusion for clinical events.
Genera associated with pathogenicity in conditions of immune exhaustion (Corynebacterium, Lautropia) were predominant in patients with sepsis. Metabolic pathways associated with oxidative stress and endotoxemia [lipopolysaccharide (LPS) synthesis and sulfur relay] were significantly upregulated in sepsis. Specific taxa were associated with sites of infection in DC patients. Protective oxidant pathways that increase glutathione were upregulated in those without sepsis. Gammaproteobacteria family of sulfur-metabolizing bacteria, exaggeration of orally predominant pathogens (Prevotella), and pathways of severe LPS-related hyperinflammatory stress were notable in those with interleukin-6 levels >1,000 pg/dL. Pathogenic genera related to an immune deficient state was significant in DC with ≥2 infection episodes. Megamonas was associated with survival during the same admission.
Specific gut microbiota and their metabolites were associated with sepsis and related events in patients with DC. Identifying beneficial strains that reduce immune exhaustion and supplementation of favorable metabolites could improve therapeutics for DC and sepsis, for which larger prospective, well controlled population-based studies remain an unmet need.
The human microbiome, a collective genome of millions of bacteria, viruses, and fungi, have a sophisticated, multidirectional, and mutualistic relationship with their human host. At the core of this interaction is the gastrointestinal tract, which contains trillions of bacteria that create a complex ecosystem called the bacterial gut microbiota (BGM). The BGM affects host health and play major role in acute as well as chronic disease conditions. Autochthonous BGM impede pathogens through selective and controlled expansion, beneficial metabolite and nutrient production, and endocrine interactions while closely corresponding with the local and systemic immune system. Disruption in these mechanisms interact with other factors such as environmental (occupation and drug/toxin exposure), host (genetic predisposition, alcohol, and tobacco), and disease states (metabolic syndrome and cirrhosis). That leads to initiation of both quantitative (diversity reduction) and qualitative (perturbed functional metabolism) changes in the BGM, a process known as dysbiosis. Dysbiosis worsens existing disease (e.g., cirrhosis progression) or causes new clinical events (e.g., infections).1 The BGM plays a central role in the etiology and progression of various liver diseases (e.g., alcohol associated hepatitis) and clinical events in cirrhosis (e.g., hepatic encephalopathy), and its modulation ameliorates adverse patient outcomes.2–4 Recent evidence had demonstrated that dysbiosis of the BGM and its functions promote sepsis, “a life-threatening organ dysfunction caused by dysregulated host response to infection,” and associated organ dysfunction in affected patients. New research has shown that microbiota modulation may help improve clinical outcomes and organ dysfunction associated with sepsis.5,6 Nonetheless, the role and function of BGM in patients with advanced cirrhosis and sepsis remain unknown. The current work therefore aimed to characterize the BGM and its functions associated with specific clinical and biochemical events among decompensated cirrhosis (DC) patients who were infected (iDC) or noninfected (niDC).
The study objectives were to characterize significant bacterial communities and their functional metabolism between iDC and niDC patients and distinguish and identify significant bacterial groups and associated functional metabolites among iDC patients grouped into the following sepsis specific categories: (1) infection sites, (2) infection episodes (0, 1, or ≥2), (3) Systemic inflammation in sepsis evaluated by interleukin (IL)-6 (<100, 100–1,000, and >1,000 pg/dL), (4) outcome during hospital stay, and (4) outcome 180 days after admission. We also described the topology of the relationships between pertinent bacterial taxa associated with grouped parameters and events in DC with and without infections using network analysis.
From March 2019 to December 2019, consecutive patients with decompensated cirrhosis admitted to the Liver Unit through the emergency and outpatient departments were included (Fig. 1). This study protocol was approved by the Institutional (hospital) review board and was performed in accordance with the Helsinki declaration of 1975 and its revisions. All participants provided informed consent for the use of de-identified fecal samples. A comprehensive clinical and drug history was obtained at admission, including events occurring over the preceding 3 months. This included details on prior admission for infection, exhaustive evaluation during the previous admission and discharge summaries when available, and decompensation events. Decompensation was defined as the presence of either jaundice, ascites, kidney injury, hepatic encephalopathy, or a combination of those. Patients with suspected drug-induced liver injury (including complementary and alternative drugs), those listed for liver transplantation, and those with index presentation as acute on chronic liver failure were excluded. Similarly, patients with acute variceal bleeding, those with hepatic and extrahepatic malignancies, those undergoing interventional vascular and invasive hepatobiliary procedures, those with severe cardiopulmonary disease, those with advanced cirrhosis on multiple organ support, and those patients on hemodialysis were also excluded. Infection was suspected based on history and clinical and physical examination and confirmed either by radiological, biochemical, or microbiological evidence along with blood and body fluid analyses. All patients with serum procalcitonin ≥10 ng/mL were considered to harbor bacterial infection even without an identifiable source. When serum procalcitonin was <10 ng/mL infection was confirmed when either radiological, blood and body fluid, or microbiological evidence supported the same.7,8 IL6 levels (normal <7 pg/mL) were documented only to determine systemic inflammation and not to identify infection or sepsis. Decompensated cirrhosis patients with and without infections were matched for the presence of ascites, hepatic encephalopathy (HE), and acute kidney injury, Child Turcotte Pugh stage, and model for end-stage liver disease scores at baseline using the case–control matching function in MedCalc software (Ostend, Belgium). All included patients were started on intravenous third-generation cephalosporins, and antibiotics were modified per site of infection and culture sensitivity during the course of hospitalization.
Stool samples were collected within 24 h of hospital admission in sterile containers, immediately processed into aliquots within 1 h, and stored at −80°C in the in-house storage designated for the protocol. After collecting a minimum of 10 samples during a period, RNAlater (Ambion/Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) was added to the aliquots, which were then transported to the main laboratory facility within 4 h while maintaining a cold chain −20°C for DNA extraction and further analysis.
Approximately 200 mg of the provided stool sample was used for bacterial DNA extraction, using a defined and published protocol modification of the commercially available QIAmp DNA Stool Mini Kit1 (Qiagen, Venlo, Netherlands). Sequencing as per standardized validated methods at the V3-V4 regions was performed with an Illumina MiSeq next-generation sequencer (Illumina, San Diego, CA, USA) using the Illumina kit at 2×300 paired-end sequencing, after which taxonomic classification was performed according to the GreenGenes Database (version 13.8).
The alpha diversity, including phylogeny (i.e. summed evolutionary age of all the species in the community), was presented using phylogenetic diversity (PD or Faith’s PD) measure, which uses phylogenetic distance to calculate the diversity of a given sample. Phylogenetic distance represents the number of changes that have occurred within a particular taxon (branch) ascertained within groups or between groups. Quantitative Insights into Microbial Ecology (QIIME version 2) was used to ascertain the quantitative and qualitative microbial communities.9 The Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway was utilized to study functional metabolic interactions and relationships within sequenced bacterial communities.10 Analysis of Similarity was used to test for a statistical difference between two or more microbial communities (alpha diversity). A p-value <0.05 was considered to indicate statistically significant differences between grouped sample attributes.11 The visualization tool Circos (v.0.69-9; tool version: 0.23) was used to facilitate the exploration and analysis of similarities and differences arising from comparisons of bacterial communities. The table viewer script in Circos tools was used to format abundance data. This method shows significant positive, negative, or neutral interactions between communities specified for the variable. Output data were presented using a circular chord ideogram layout that displayed the relationships among microbial communities concerning clinical or investigation variables at the phylum/class/order/family taxonomic level. To simplify the graphical output, tabulations were made by compounding all analyzed clinical or investigation variables.12 Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt, version 1.1.1) was used for predictive metabolic functional profiling of microbial communities using 16S rRNA marker gene sequences precalculated for protein-coding genes present in KEGG gene families and 16S rRNA gene copy number.13 Linear discriminant analysis effect size (LEfSe) combined with the Kruskal-Wallis and pairwise Wilcoxon tests were utilized to identify significant differences in abundance and functionality of microbial communities between groups. We used default significance (alpha value=0.05) and linear discriminant analysis thresholds (2.0) at all taxonomic levels between time points. To simplify the graphical output, tabulations on significant taxa and the associated functional metabolic pathways were provided.14
Networks were inferred using CoNet (v.1.1.1-beta) application within Cytoscape (v.3.7.2). The following measures, implemented by CoNet, were used to detect copresence/exclusion between bacterial communities: Pearson, Spearman, Mutual Information, Bray-Curtis dissimilarity, and Kullback Leibler dissimilarity. CoNet was also set up to determine the comparison metadata for each sample (i.e., the respective clinical or investigational variable grouping). Moreover, the following network measures were calculated using NetworkX (version 2.2): degree centrality, betweenness centrality, closeness centrality, and added to the node attributes in the graph. The centrality measure of degree assigns importance to a specific node (microbe at bacterial phylum/family/genus level) in the form of the number of direct connections to other nodes (microbes). The centrality measure of betweenness represents the number of times a microbe lies on the shortest path between other microbes, which identifies microbes that serve as the best “bridges” between other microbes in the network. This measure demonstrates the bacteria that influence flow around a system (in this case, the clinical or investigation variable). A high betweenness indicated the microbe that holds authority over different clusters in the specified network. The centrality measure of closeness represented microbial (nodal) closeness to all other nodes (microbes) in the network. These measure the shortest path between all nodes (microbes) and identifies the individual microbe best placed to influence the entire network more quickly or those who are good “broadcasters” of information (metabolic-cross talk). The topology of constructed networks and significant taxa, the top 2% of all taxa analyzed, were presented in a simplified tabular form based on these attribute indices of degree, closeness, and betweenness centrality measures.15–17 General statistical analysis was performed using MedCalc version 19.7 (Ostend, Belgium) and NCSS version 12 (Kaysville, UT, USA) statistical software. Data were reported as means and standard deviation or as medians and interquartile range between brackets as applicable. The Shapiro-Wilk test was used to test normality and Levene’s test was used for non-normal distributions. The Bartlett homogeneity test (for nominal variables) was utilized to assess the equality of variances. To decrease data variability and increase data conformity to a normal distribution, logarithmic transformation was applied. Chi-square and Fisher’s exact tests were used to compare normally distributed variables; the Mann-Whitney U test was used to evaluate continuous variables. A simplified summary of patient inclusion and methodology is shown in Supplementary Figure 1.
This study included 51 patients (niDC=24 and iDC=27), with both groups having an equal male distribution. Spontaneous bacterial peritonitis (SBP) was most common in patients with infections (n=8, 29.6%), followed by pneumonia and urinary tract infection (UTI, both n=5, 18.5% each), skin and soft tissue infections (SSTIs; n=4, 14.8%), spontaneous bacteremia (n=4, 14.8%), and acute cholecystitis (n=1, 3.7%). Culture positivity was noted in three patients with SBP, two with Escherichia coli (one multidrug resistant, MDR) and one Acinetobacter baumannii, four with pneumonia, three with Klebsiella pneumoniae (two MDR) and one with Staphylococcus aureus, and three with Escherichia coli UTIs (two MDR) and one pan-drug resistant; two with SSTIs (Escherichia coli and Klebsiella pneumoniae), and four with bacteremia, three Klebsiella pneumonia and one Escherichia coli (MDR). The iDC group had a significantly higher number of patients with at least one or two or more documented infection episodes within the preceding 3 months. At admission, the iDC group had significantly higher IL6 levels compared with the niDC group (median 85.7 vs. 23.5 mg/mL respectively, p<0.001). Moreover, the proportion of patients receiving an injectable cephalosporin (iDC 48.7% vs. niDC 4.1%; p<0.001), piperacillin-tazobactum (22.2% vs. 4.1%; p<0.001), or carbapenem (18.5% vs. 0%; p<0.001) within the preceding 90 days was significantly higher in the iDC group than in the niDC group despite similar rifaximin use in both groups. A significantly higher proportion of patients with active infection compared to those without infection at admission died during the same admission (iDC 29.6% vs. niDC 4.1%; p<0.001) but not at the end of follow-up. Pertinent baseline characteristics of both study groups are summarized in Table 1.
Using the QIIME (v.2) and Greengenes microbial gene database, our analysis showed significant differences in baseline alpha and beta diversities and relative abundances (RA) of bacteria at the phylum, family, and genus levels between the niDC and iDC groups. Alpha diversity differed significantly between the niDC and iDC groups, regardless of infection site, between those with no infections (niDC) and those with two or more episodes of infection (iDC) and among all interleukin cutoffs (iDC). No significant difference in alpha diversity was seen among the iDC group patients who survived or died during same admission or on overall follow-up (Supplementary Figs. 2 and 3). QIIME and Circos analysis showed that the RA of specific bacterial taxa were associated with clinical and investigation variables (Fig. 2 and Supplementary Table 1). Briefly, RAs of the beneficial butyrate-producing family Lachnospiraceae; “nitrogen fixer” family Acetobacteraceae; and the Ruminococcaceae family were found to be inversely correlated with intestinal permeability in niDC patients. Meanwhile, the predominantly pathogenic family Enterobacteriaceae, the oxidative stress-associated manganese oxidizing family Shewanellaceae, and the proinflammatory activity-associated family Oxalobacteraceae were specifically abundant in patients with infections at admission. Among the DC patients with repeated infections, opportunistic pathogens (generally skin predominant) associated with antimicrobial use-related expansion, such as Dermabacteraceae and Anaeroplasmataceae, were observed. Deviant, opportunistic, nonautochthonous phyla, such as Chromatiaceae, Brevibacteriaceae, and Intrasporangiaceae, were notably greater in patients with bacteremia compared to those with other sites of infection.
After identifying the RAs of bacterial population and key bacterial families associated with specific clinical and investigation events, the multivariate LEfSe method utilizing standard tests for statistical significance with additional tests encoding biological consistency and effect relevance was utilized to determine significant bacteria and respective metabolic functions most likely to explain the differences between grouped variables. Briefly, genera associated with “gut barrier health,” such as Bifidobacterium (antimicrobial peptide synthesis, vitamin metabolism, and beneficial butyrate production via cross-feeding mechanism) and Coriobacterium (bile salt and steroid conversion and dietary polyphenol activation), were predominant in patients without infections. Metabolic pathways associated with negating oxidative stress (selenocompound, adipocytokine signaling, and cysteine and methionine metabolism pathways) were significantly upregulated in DC patients without infection at admission. In contrast, pathogenic genera (e.g., Leptotrichia, Lautropia, Neisseria, and Acinetobacter) and metabolic pathways associated with oxidative stress and gut barrier dysfunction (e.g., lipopolysaccharide biosynthesis, bacterial chemotaxis, and sulfur relay system) were significantly expressed in DC patients with infections at admission. Among infected patients with IL6 levels <100 and >1,000 pg/mL, Propionibacterium (with immunomodulatory properties, which has a role in human health by occupying niches that are colonized by other more pathogenic microorganisms) and Paraprevotella and Caldiserica, which participate in human disease by promoting chronic inflammation, were significantly increased. This was also associated with the upregulation of propanoate metabolism associated with beneficial short chain fatty acid (SCFA) generation in the former and pentose phosphate pathway associated with heightened oxidative stress in the latter. Similar specific bacterial taxa and metabolic pathways also differed significantly according to number, type, and sites of infection. DC patients who survived sepsis-related admission and were alive until the end of follow-up had significantly more Megamonas at baseline, whereas those who died on follow-up were Kingella and Neisseria predominant (Fig. 3 and Table 2).
Network analysis for interactions using NetworkX revealed striking differences and interactions between bacterial communities when grouped according to clinical and investigational parameters. The main advantage of network topology analysis is that it allows us to understand the type of interaction or association between bacterial communities. Accordingly, cirrhosis patients with and without infections had lower interactions of mutual exclusion (negative association) and higher interactions of co-occurrence (positive) at baseline, respectively (Fig. 4A). Similarly, the core bacterial taxon (at the family level) influencing other bacterial communities differed when grouping cirrhosis patients according to episodes of infection: Erysipelotrichaceae in those without infection, Verrumicrobiaceae in those with a single infection, and Ruminococcaceae in those with two or more episodes of infection (Fig. 4B). Further network analysis using circular attribute function showed that various bacterial taxa at the phyla and family levels interacted with each other differentially in cirrhosis patients when grouped according to presence or absence of infection, episodes of infection, and severity of systemic inflammation based on IL6 levels (Fig. 4C, D). After analyzing the centrality measures of degree (interaction within the network), betweenness (influence within the network), and closeness (broadcast within the network), interesting changes were observed in the various topological attributes of the constructed networks among groups with respect to different grouping attributes. For instance, Blautia was most abundant in patients with sepsis, whereas Gemella was most interactive, Leptotrichia was most influential, and Lactobacillus ruminis was most central to the metabolic-cross talk within the network. Similarly, in patients with lung infection, Klebsiella pneumoniae was most abundant, whereas Streptococcus was most interactive and Bifidobacterium longum and Enterococcus were most influential and central to crosstalk, respectively. Meanwhile, in those with bacterial peritonitis, Bifidobacterium, Leptotrichia, Streptococcus, and Lactobacillus was most abundant, interactive, influential, and central to crosstalk, respectively. Supplementary Table 2 depicts the network topology in a simplified tabular form for comparison between various groups, based on attribute indices of degree, closeness, and betweenness centrality measures.
This study defined and demonstrated the quantitative, qualitative, and interactive roles of the BGM in the presence and absence of infections. To the best of our knowledge, no study has comprehensively analyzed and characterized the BGM with respect to clinical events associated with infections in cirrhosis. Our study aimed to provide insights on the role of BGM in promoting sepsis and associated events among patients with cirrhosis. Accordingly, our findings showed that pathogenic genera, such as Leptotrichia, Neisseria, and Erwinia, were predominant in cirrhosis patients with infection, whereas beneficial bacteria, such as Bifidobacterium and Coriobacteria, were predominant in those without infection. Similarly, the functional metabolism of the predominant BGM included upregulation of pathways associated with oxidative stress, endotoxemia, and proinflammatory process in those with infections and endogenous antimicrobial, antioxidant, and cytokine signaling pathways in those without infections. Specific pathogenic bacterial taxa were associated with different sites of sepsis (e.g., Mycoplasmataceae in bacteremia and Aeromonas in UTIs). Repeat infections and high systemic inflammation were associated with rare, multidrug-resistant, and immune exhaustion-related intestinal bacterial groups. This change in BGM was significantly notable in those without infections (Bifidobacterium) who subsequently developed their first (Lautropia) and multiple episodes of infection (Leptotrichia). Specific bacterial genera with high propanoate functional metabolism, such as Megamonas, were associated with survival in infected patients with cirrhosis during admission and after follow-up. Network analysis showed that noninfected cirrhosis patients had richer and more diverse interactions, comprising mutual exclusions and co-occurrence between various bacterial taxa, compared with those with infections. Host-microbiota interactions can predispose individuals to infections, promote more severe infections, and even prevent their occurrence, as demonstrated in insects, small animals, and humans. Microbe-arbitrated host protection from infections can occur through the maintenance of host immune homeostasis, interference competition with pathogenic taxa, and competitive resource utilization. A decline in host health, as seen in development and progression of cirrhosis, can cause dysbiosis, which promotes the perturbation and transition of commensal microbiota toward pathogenicity and further predisposes the host to infections.18 Reduced microbial diversity and loss of colonization resistance due to perturbation caused by dysregulated local and systemic immune homeostasis promote systemic infections originating in the gut, as demonstrated with Clostridioides difficile infection.19 Gut microbiota can be used to develop a pathogenic profile with the onset of dysbiosis driven by illness and its management or immune compromise, such as antibiotic treatments and liver disease, respectively.20,21 Liu et al.22 found that changes in gut microbial patterns promoted necrotizing enterocolitis and late-onset sepsis in babies born prematurely and that changes in gut bacteria were probably the causative factors of the development of infectious complications in predisposed patient groups. Moreover, changes in gut microbiota composition can potentially predispose patients to a state of immunosuppression, thereby increasing the risk of sepsis. Indeed, Hyoju et al.23 clearly demonstrate how a selective high-fat, sucrose-rich Western diet, antibiotic exposure, and surgical injury converge on the microbiome, causing lethal sepsis and multiorgan failure without exogenous pathogens. In immunosuppressed patients, such as those undergoing bone marrow transplantation, antibiotic-associated dysbiosis promoted a five- to nine-fold increase in the risk of bloodstream infection and sepsis. Similarly, a strong dose-response relationship between dysbiosis-causing events and subsequent severe sepsis-related hospitalization had been observed among elderly patients. Furthermore, a previous study showed that prolonged antibiotic exposure and utilization of additional antibiotic classes and broader-spectrum antibiotics during hospitalization were associated with dose-dependent increases in the risk of subsequent sepsis. This suggested that the association between antibiotic exposure and subsequent sepsis was associated with microbiome depletion and not illness severity.24–26 Critical illness, as noted in patients with decompensated cirrhosis, profoundly disturbs the gut microbiota due to low food intake and malnutrition, recurrent hospitalizations for portal hypertensive complications, worsening liver function and progressive decline in immune status, and multiple exposure to antibiotics. This phenomenon is evident in the current study, wherein patient with advanced cirrhosis requiring multiple admissions demonstrated worsening dysbiosis due to repeated infections, causing pathobiont generation, immune exhaustion, multiple organ failure, and poor survival. The effects of the gut microbiome on sepsis outcome have been clearly demonstrated in an animal model. Notably, when sepsis was introduced using the cecal ligation and puncture method in two different sets of mice purchased from different vendors, with different fecal microbiota beta diversities and immune phenotypes, higher mortality was notable in the first group. When both groups of mice were housed together, differences between beta diversities and immune phenotypes disappeared, and further sepsis introduction included similar survival outcomes. The aforementioned study highlighted the importance of the gut microbiome for survival from, and host immune response to, sepsis. Thus, the transition of a microbiome into a pathobiome could promote poor clinical outcomes and mortality from sepsis by modulating the host immune response, which was seen in our patients with repeated infections, severe unchecked inflammation, and higher mortality.27 A prior study that evaluated the effects of gut microbiome dynamics in intensive care unit (ICU) patients showed that changes in the gut microbiota were associated with patient prognosis. The proportions of Bacteroidetes and Firmicutes significantly changed during ICU stay, and extreme changes were observed in almost all patients with a poor prognosis, suggesting a correlation between qualitative and quantitative dysbiosis and sepsis outcomes. Similarly, a study conducted by our group demonstrated that progressive dysbiosis in advanced cirrhosis with repeated infections and hospitalizations promoted extreme changes in fecal bacterial communities and was associated with poor short- and long-term clinical outcomes.28 Specific classes of antibiotics have been associated with fecal microbiota changes. Accordingly, penicillins, cephalosporins, and carbapenems were associated with a reduction in beneficial Bifidobacteria and Lactobacilli and an increase in pathogenic Enterobacteriaceae/Enterococci. Published faucal microbiota data suggest the development of resistance in Enterobacteriaceae following exposure to piperacillin-tazobactam. Moreover, antibiotic-associated microbiota disruption can occur as early as 3 days and last for 12 months as shown via molecular analytical methods in those receiving short courses of fluroquinolones.29 This study found that infections upon admission in DC patients were associated with deleterious changes in the fecal microbiome, which progressively worsened in those with repeated infections. The GBM of those receiving high-end antibiotics for worsening infections transitioned toward a pathobiont profile, and associated bacterial taxa were identified as opportunistic pathogens predominant in immunocompromised states. Prior studies on patients with and without sepsis demonstrated heterogeneous patterns of intestinal microbiota, including the disappearance of bacterial genera with important functions in host metabolism. Moreover, our fecal analysis of critically ill patients with sepsis at different sites revealed significant differences in GBM profiles, similar to our findings among patients with cirrhosis.30 Agudelo-Ochoa et al.31 found that the microbiota of ICU patients with sepsis contained numerous microbes strongly associated with inflammation, such as Parabacteroides, Fusobacterium, and Bilophila species. Furthermore, a difference in the abundance of pathogenic species, such as Enterococcus spp., was observed in patients with sepsis who died. Among our patients, those with infections, those with repeated infection, and those who survived after matching for portal hypertension events, liver disease severity, and extrahepatic organ failure involvement, had a GBM that was preferentially and significantly associated with proinflammation, intestinal barrier dysfunction, and endotoxemia. We chose to study IL6 in our patients for several reasons. Studies have shown that IL6 levels are increased in patients with cirrhosis and is linearly associated with the severity of underlying cirrhosis. IL6 was also found to play a role in the hyperdynamic circulation observed in patients with cirrhosis. In DC, plasma IL6 levels on admission were shown to provide the most sensitive and specific tool for the diagnosis of bacterial infection and were closely linked to the development of overt HE in patients with liver cirrhosis and was particularly useful in predicting mortality in patients with cirrhosis.32–35 In that context, our study revealed the important role of GM dysbiosis and specific bacterial taxa associated with varying IL6 levels in DC patients with infections. Short-chain fatty acid-producing Propionibacterium was relatively abundant in those with IL6 <100 and pathogenic taxa were associated with states of immune exhaustion and opportunistic infections predominated the gut in DC patients with IL6 >1,000. That demonstrated a plausible association of GM as a promoter and driver of systemic inflammation and eventually organ dysfunction in DC patients with sepsis. This study is not without limitations. Apart from being a single-center, retrospective study, we did not match our patients in terms of other confounders that could have affected gut microbiome, as well as long-term outcomes, such as dietary habits, heterogenous environmental exposures, and other known or unknown chronic comorbidities. Nonetheless, the DC groups with and without infections were comparable with respect to liver disease severity, portal hypertension events, nonabsorbable antibiotic use, and extrahepatic organ dysfunction at admission. We did not assess dynamic changes in the BGM from admission to discharge or the end of follow-up, given that it could have been affected by heterogenous and possibly unmodifiable confounding factors. During follow-up, we did not account for the various causes of death in our patients. Further, the generalized clinical outcomes, such as survival or death, were based on a single point fecal microbiota analysis for ease in analysis, which could have oversimplified the role of gut dysbiosis in our patient population. Among patients with cirrhosis, dysbiosis occurs as part of the etiology and development of liver disease and subsequently promotes infections in advanced cirrhosis due to portal hypertension and immune dysfunction. That worsens with subsequent hospitalization and therapeutic or symptom management interventions required in this special patient group, which prevents the complete recovery of the microbiota. Periods of dysbiosis predisposed DC patients to opportunistic infections, which further worsened dysbiosis, and predisposed them to sepsis, wherein the microbiota composition becomes extremely compromised and the risk of secondary infections, immunosuppression, and organ dysfunction increases. The study showed that gut microbiota changes in advanced DC patients induced a state of prolonged immunosuppression, rehospitalization because of infections, and increased mortality. Similar to the feasibility and effectiveness of fecal microbiota transplantation as a treatment option for severe C. difficile infection in the ICU, the findings suggest that gut microbiota modulation using healthy stool transplants could be an important therapeutic option in the fight against multidrug resistant bacteria in vulnerable patient populations, such as those with advanced cirrhosis, pending further large quality, well controlled observations and clinical trials. Future directions based on the current work includes defining subtypes of patients (independent from other confounders) based on the microbial signatures or metabolites that would benefit from targeted therapies; validation of our findings in small animal models or a second patient cohort (with a different geographic or ethnic background) prospectively, and analyzing longitudinal data (e.g., after antibiotic therapy, in patients with antibiotic prophylaxis for infections such as SBP).
Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. |
|
PMC9647107 | Yimou Lin,Haitao Huang,Lifeng Chen,Ruihan Chen,Jimin Liu,Shusen Zheng,Qi Ling | Assessing Donor Liver Quality and Restoring Graft Function in the Era of Extended Criteria Donors | 16-08-2022 | Liver transplantation,Extended criteria donors,Graft quality,Assessment,Biomarkers,Restoration | Liver transplantation (LT) is the final treatment option for patients with end-stage liver disease. The increasing donor shortage results in the wide usage of grafts from extended criteria donors across the world. Using such grafts is associated with the elevated incidences of post-transplant complications including initial nonfunction and ischemic biliary tract diseases, which significantly reduce recipient survival. Although several clinical factors have been demonstrated to impact donor liver quality, accurate, comprehensive, and effective assessment systems to guide decision-making for organ usage, restoration or discard are lacking. In addition, the development of biochemical technologies and bioinformatic analysis in recent years helps us better understand graft injury during the perioperative period and find potential ways to restore graft function. Moreover, such advances reveal the molecular profiles of grafts or perfusate that are susceptible to poor graft function and provide insight into finding novel biomarkers for graft quality assessment. Focusing on donors and grafts, we updated potential biomarkers in donor blood, liver tissue, or perfusates that predict graft quality following LT, and summarized strategies for restoring graft function in the era of extended criteria donors. In this review, we also discuss the advantages and drawbacks of these potential biomarkers and offer suggestions for future research. | Assessing Donor Liver Quality and Restoring Graft Function in the Era of Extended Criteria Donors
Liver transplantation (LT) is the final treatment option for patients with end-stage liver disease. The increasing donor shortage results in the wide usage of grafts from extended criteria donors across the world. Using such grafts is associated with the elevated incidences of post-transplant complications including initial nonfunction and ischemic biliary tract diseases, which significantly reduce recipient survival. Although several clinical factors have been demonstrated to impact donor liver quality, accurate, comprehensive, and effective assessment systems to guide decision-making for organ usage, restoration or discard are lacking. In addition, the development of biochemical technologies and bioinformatic analysis in recent years helps us better understand graft injury during the perioperative period and find potential ways to restore graft function. Moreover, such advances reveal the molecular profiles of grafts or perfusate that are susceptible to poor graft function and provide insight into finding novel biomarkers for graft quality assessment. Focusing on donors and grafts, we updated potential biomarkers in donor blood, liver tissue, or perfusates that predict graft quality following LT, and summarized strategies for restoring graft function in the era of extended criteria donors. In this review, we also discuss the advantages and drawbacks of these potential biomarkers and offer suggestions for future research.
Liver transplantation (LT) is a life-saving treatment option for patients with end-stage liver disease. In recent decades, good short- and long-term outcomes after LT have been achieved because of improvements in surgical technologies and organ preservation.1 Graft quality is believed to play a dominant role in early graft function and thereby dramatically influences graft survival and mortality after LT.2–4 Over the last decade, the disparity between the need for LT and the organ shortage is widening, which leads to the expanded usage of grafts from the extended criteria donors (ECDs).1 Traditionally, ECDs are donors with underlying medical diseases such as diabetes, or hypertension, advanced age, high-degree liver steatosis, prolonged ischemia time, pathogenic infection, prolonged intensive care unit stay, hypernatremia, and donation after circulatory death (DCD).5–7 ECD graft quality is routinely considered inferior because of their increased rate of post-transplant complications, such as primary graft nonfunction (PNF),2 early allograft dysfunction (EAD),4 and ischemic-type biliary lesions (ITBLs)8,9 PNF is early graft loss after LT and requires emergency regrafting, which occurs following 2–10% of LTs.10–12 ECDs include DCD donors13 and those with severe steatosis,14 prolonged ischemia time,15–17 and high donor bilirubin level18 sharply increase the risk of PNF, thereby reducing patient and graft survival. Unlike PNF, EAD represents marginal, usually reversible, graft function during the first postoperative week, and results in a higher morbidity and mortality.4 Compared with 1–10% seen in donation after brain death (DBD) LT, the incidence of biliary complications after DCD LT is approximately 10–30%,19–22 in which the time from asystole to cross-clamp is considered as a major risk factor.23 Moreover, advanced donor age, prolonged ischemia time, microvascular thrombosis, bile salt toxicity and immune injury may be the underlying mechanisms of the development of biliary complications.24,25 Therefore, ECDs should be well defined and precisely allocated to appropriate recipients. More importantly, in the era of ECD, effective systems need to be established to assess donor liver quality and guide the decision for organ usage or discard. Based on clinical risk parameters (Fig. 1), models like donor risk index,2 Eurotransplant donor risk index,26 and discard risk index18 were constructed to evaluate the risk of graft failure or discard, serving as useful tools to make decisions for organ allocation.2,26 However, those scoring models mainly focus on donor characteristics and cannot assess the degree of liver injury.27 Furthermore, combining clinical parameters with advanced molecular profiles, imaging, or histopathology may contribute to the development of better systems. In recent years, with the rapid development of multi-omics, single cell technology, and bioinformatic analysis, significant achievements have been made in revealing the molecular profiles that are closely related to poor graft outcomes, and which can provide novel biomarkers for evaluation of graft viability. Herein, we provide a review of potentially useful biomarkers in donor blood, liver tissue, and graft perfusate, which have been associated with impaired graft quality or predictive for the occurrence of EAD, PNF, and biliary complications after LT. In this review, we mainly focus on studies using human liver grafts. Given that the available biomarkers were insufficient in the field of LT, we also include experimental studies that have been performed in animal models. Furthermore, we summarize potential therapies for graft repairment during LT. Finally, we describe the pros and cons of the potential biomarkers, accompanied with suggestions for future graft assessment and restoration.
Donor serum alanine transferase (ALT), aspartate transferase (AST), total bilirubin, gamma glutamyl transpeptidase, and sodium concentration may reveal the underlying liver dysfunction and ischemic injury prior to graft procurement. Over the past decades, numerous studies have demonstrated that such laboratory disorders in donor blood are independent risk factors for early graft dysfunction following LT.18,26,28 In recent years, novel biomarkers in donor blood have been found to useful for predicting graft outcomes. By analyzing data from over 10,000 nondiabetic donors, Ezekian et al.29 showed that elevated donor serum hemoglobin A1c (HbA1c) >6.5% was associated with increased rate of PNF and decreased graft and patient survival. HbA1c is known to be a useful biomarker representing the average plasma glucose concentration within the last 3 months, serving as an early warning of diabetes. The liver undergoes glycogen deposition and hepatic steatosis resulting from diabetes.30,31 Therefore, it is worth noting that HbA1c may be a valuable marker for further stratifying marginal graft quality. In a large prospective study of 815 participants, Piemonti et al.32 identified increased serum donor interleukin 6 (IL6) and C-X-C motif chemokine ligand 10 (CXCL10) concentration as predictors of poor early graft function, graft failure and inferior graft survival after DBD LT. IL6 is responsible for transforming naïve B cells into mature plasma cells, as well as activating the production of IL17 to inhibit regulatory T lymphocyte (Treg) function.32 Alternatively, CXCL10 is a useful chemoattractant for macrophages, natural killer (NK) cells and dendritic cells (DCs), thereby shaping initial immunity.32 More interestingly, Pollara et al.33 found that elevated circulating mitochondria-derived damage-associated molecular patterns (mtDAMPs) in donor plasma were associated with severe inflammation response and the development of EAD following DBD LT in a group of 55 recipients. The major source of mtDAMPs may be the mitochondria released from graft tissue or cell death during organ procurement, suggesting that mtDAMPs might quantitatively assess graft injury.
The liver, a multifunctional organ in the body, is mainly engaged in metabolism, synthesis, storage, detoxification, and complex immune activities. After implantation, the donor graft becomes the new center of the recipient to perform those functions.34 Therefore, the graft features could significantly regulate hepatic homeostasis and influence outcomes after LT (Table 1).35–55 Donor grafts could be gained for histological assessment and quantification of liver injury during LT. Histopathology is the gold standard for the diagnosis of steatosis, fibrosis, necrosis, inflammation, and cellular infiltration in liver grafts. In our center, pretransplant, and post-reperfusion liver biopsies are routinely performed, offering valuable clues for graft quality assessment (Supplementary Table 1).56 In addition, bile duct biopsies could provide valuable information to evaluate bile duct injury and predict graft outcomes. Dries et al.57 proposed a scoring system (Supplementary Table 2), including biliary epithelium, mural stroma, peribiliary vascular plexus, thrombosis, intramural bleeding, peribiliary gland, and inflammation, to quantify bile duct injury.
With the advent of genome-wide association studies and pretransplant genetic analysis, a series of genes and variants have been found to be susceptible to graft injury.58 Heme oxygenase-1 (HO-1), a regulator of immune response, is considered to be cytoprotective gene of ischemia-reperfusion injury (IRI) during LT and is modulated by a single-nucleotide polymorphism A (-413) T.35 Buis et al.35 reported that, compared with recipients of a liver with an A-allele genotype (n=245), recipients of livers with an HO-1 TT-genotype (n=61) had dramatically elevated serum hepatic transaminases after LT and a higher incidence of PNF. HLA-C, which is the major inhibitory ligand for immunoglobulin-like receptors, inhibit the cytotoxic activity of NK cells, and therefore reduced liver inflammatory damage.59 In a large LT cohort of 459 patients, Hanvesakul et al.36 found that donor grafts with at least one HLA-C2 allele were associated with less incidence of graft dysfunction and rejection. After LT, graft-derived cell-free DNA (GcfDNA), which is continuously released into recipient circulation because of cellular turnover, is a promising noninvasive biomarker to assess graft quality. Previous studies have showed that the elevated GcfDNA was a signal of early graft injury after LT, particularly acute cellular rejection.37,60,61,38 For example, a prospective study conducted by Schutz et al.38 demonstrated that GcfDNA increased by more than 50% 1 day following LT, probably because of the IRI. However, GcfDNA rapidly decreased to a median of <10% within 7–10 days without the recipient experiencing early graft injury over a 1 year observation period.38 This suggested that GcfDNA may be a precise and superior biomarker to predict early graft dysfunction compared with conventional liver function tests.
Protein-coding associated RNAs, for example messenger RNA (mRNA) and noncoding RNAs including microRNAs (miRNAs), circular RNAs (circRNAs) and long noncoding RNAs (lncRNAs) are believed to be reliable markers to evaluate graft injury because of their organ specificity. Nrf2 transcription factor, which is activated by reactive oxygen species, is known to protector against liver IRI via activating phase II antioxidants.62 Zaman et al.39 demonstrated that grafts (n=6) with increased Nrf2 mRNA expression before IRI were associated with lower liver injury. Interestingly, donors with low Nrf2 mRNA levels (n=8) were significantly older than those with high levels, suggesting that older grafts experienced severe IRI39 and inferior graft quality. Additionally, Resch et al.40 reported that high gene expression of the major histocompatibility complex class 1 related chain A (MICA) mRNA in zero hour biopsies (n=88) was associated with mild graft injury and prolonged graft survival. During LT, MICA had an important role in linking the innate and adaptive immune responses via interacting with NK cells, mucosal-associated invariant T, CD8+T cells, et al.40 miR-22, a regulator of a series of pathways such as cell cycle, metabolism and kinase signaling, is relevant to cell survival, glucose metabolism, and protein translation.41 Khorsandi et al.41 rereported that low expression of graft miR-22 was associated with the incidence of PNF after DCD LT (n=21). Another study of 42 human LTs showed that high expression of donor graft miR-146b-5p was associated with the development of EAD.42 Downregulation of miR-146b increased the production of tumor necrosis factor receptor-associated factor 6, which activated the nuclear factor-kappa B (NF-κB) pathway, and in turn enhanced Treg function.42,63 In our previous study, we found that elevated donor graft miR-103 and miR-181 were significantly associated with the development of new-onset diabetes mellitus (NODM) in recipients following LT (n=30).43 NODM not only increased the risk of biliary stricture and cholangitis but also resulted in poor graft survival, serving as an indicator of poor graft quality as well.64 The two miRNAs targeted several genes related to glucose homeostasis and insulin signal transduction, which may have been the underlying mechanism.43 In a cohort of 115 human LTs, Wang et al.44 reported that low levels of donor graft circFOXN2 and circNEXTIN3 that regulated miR-135b-5p and miR-149-5p and had roles in hepatic IRI were associated with the incidence of EAD. In a mice model of IRI (Qu et al.65 identified 13 differentially expressed circRNAs (e.g., Chr3:83031528|83031748, Chr10:89473752|89483524) in postperfusion livers that were involved in more severe IRI in steatotic livers. In a rat LT model, Chen et al.45 demonstrated that lncRNA LOC103692832 in rat grafts was related to early graft injury following LT that was mediated by the expression of apoptosis-related genes like HMOX1 and ATF3. Nevertheless, the mechanisms of these potentially involved circRNAs and lncRNAs are still unclear, and further prospective or multicenter studies with larger samples are needed to verify the results.
Sirtuin1, a histone/protein deacetylase that regulates inflammatory responses, cellular aging, and stress resistance, has an important role in autophagy induction involved in liver IRI.66 A previous study showed that high Sirtuin1 expression in grafts post-reperfusion sharply inhibited proinflammatory cytokine levels accompanied by superior liver function and improved patient survival.46 HO-1 is a rate-limiting enzyme that converts heme to biliverdin, free iron, carbon monoxide, and has anti-inflammatory and anti-oxidative activitiy.47 In addition, Nakamura et al.47 showed that high HO-1 levels in post-reperfusion liver biopsies (n=51) were associated with good liver function, dramatically enhanced Sirtuin1/LC3B expression, and protected against hepatic IRI by inducing autophagy. Notch1, a highly conserved transmembrane receptor, has been shown to reduce cellular apoptosis or necrosis and inflammatory response.55 Kageyama et al.44 demonstrated high Notch1 expression in grafts was correlated with low serum ALT levels, consistent with alleviated liver damage. In addition, Liu et al.48 found that high graft YAP expression after LT was linked with well-preserved histopathology and improved liver function at 1–7 days following LT. YAP is an effector of Hippo pathway and regulates cell proliferation and apoptosis and maintains hepatic homeostasis. FGF15, which is secreted from the ileum following inflammatory stimulation, binds to Fgfr4/Klb, which is followed by downregulation of CYP7A1 expression and inhibition of bile acid synthesis and activation of the Hippo pathway to upregulate YAP levels.49 In a rat DBD LT model, Gulfo et al.49 reported ed that low FGF15 levels in grafts was associated with more severe hepatic damage and inhibited regeneration that was mediated by increased CYP7A1 and decreased YAP levels. The use of ECD grafts has raised the incidence of graft dysfunction, which ranges from reversible dysfunction, known as EAD, to irreversible dysfunction or PNF. Therefore, biomarkers to predict EAD and PNF are necessary in the era of ECD. CEACAM1 is a glycoprotein involved in hepatocyte differentiation and regeneration and regulation of insulin clearance, serving as a bridge between hepatic injury and metabolic homeostasis.50 Low CEACAM1 expression in human donor liver biopsies (n=60) was recognized as an independent predictor of EAD.50 In a large study cohort (n=194), Hirao et al.51 found that liver grafts with high occult collagen deposition were of increased risk of severe IRI and EAD, highlighting the effect of occult fibrosis on post-transplant outcome. In addition, Kurian et al.68 investigated several upregulated signaling pathways including NF-κB and targets such as CXCL1, IL1, TRAF6, TIPARP, TNFRSF1B, as predictors of EAD. Kornasiewicz et al.69 used graft proteomics to identify 21 significantly differentially expressed proteins in patients with (n=3) and without PNF (n=6). The proteins were mainly associated with mitochondrial oxidative phosphorylation or vital for the adenosine triphosphate-dependent turnover of proteins.
Cortes et al.52 used metabolomic profiling of 124 graft biopsies to identify significantly increased lysophospholipids, bile acids, phospholipids, sphingomyelins, and histidine metabolism products that were predictors for EAD. Based on the metabolic features, an EAD predictive model was established and further determined in a validation set (n=24) to have 91% sensitivity and 82% specificity. Likewise, Faitot et al.53 reported that lactate concentrations >8.3 mmol/g and phosphocholine concentrations >0.646 mmol/g were significantly associated with EAD. In our previous study, we identified metabolic profiles containing 57 dramatically differentially expressed metabolic features that were enriched in 24 common pathways including fatty acid, alanine, aspartate, thiamine, and riboflavin metabolism, the urea cycle, and ammonia recycling in PNF grafts.28 Graft metabolites and clinical characteristics were combined to develop a PNF predictive model derived from eight selected metabolic variations including achillicin, 3-hydroxypropanal, 3-oxododecanoic acid glycerides, and dopexamine in combination with clinical parameters including donor total bilirubin >2 ng/mL, graft weight >1.5 kg, cold ischemia time >10 h, graft warm ischemia time >60 m. The model had an area under curve of 0.930 for predicting PNF.28
Recent advances in single cell RNA sequencing (scRNA-Seq) allow investigation of the transcriptomic landscape of single cells in organisms and have increased our understanding of the heterogeneity and relevance between cells. In a rat LT model, Yang et al.54 identified 11 kinds of cells in grafts and drew a single cell map of IRI after steatotic LT by scRNA-Seq. More importantly, they found a pro-inflammatory phenotype of Kupffer cells (KCs) that highly expressed colony-stimulating factor 3 and a subset of DCs with high expression of XCR1 that were enriched in steatotic grafts, suggesting their participation in fatty graft IRI.54 In addition, Wang et al.55 described a dynamic transcription profile of intrahepatic cells during LT by performing scRNA-Seq of grafts at preprocurement, at the end of organ preservation, and 2 h after reperfusion. They also found that a cluster of KCs that highly expressed TNFAIP3 interacting protein 3 after reperfusion, protected grafts against liver IRI.55 We believe that as research on scRNA-Seq deepens, it may provide a deeper understanding of mechanisms related to liver IRI during LT, identify grafts at increased risk of IRI and develop strategies to protect organ against liver damage. In a study published on BioRxiv, we established a graft-tolerant mouse LT model and identified two stages of graft recovery, which included an acute and stable phases.70 We also found that the interaction between CD206+MerTK+ macrophages and CD49a+CD49b− NK cells regulated metabolic and immune remodeling of the graft.70
The donor graft and perfusate keep interplaying during preservation. Molecules including nucleic acids, proteins, and metabolites in perfusate may be associated with graft outcomes. In a review by Verhoeven et al.71 in 2014, ALT, AST, lactate dehydrogenase, lactate, adenine nucleotide level, hyaluronic acid, thrombomodulin and inflammatory markers (e.g., hypoxia-inducible factor-1α, and tumor necrosis factor-α) in perfusate and perfusate pH were useful biomarkers to assess graft quality. Machine perfusion (MP) such as hypothermic machine perfusion (HMP), hypothermic oxygenated perfusion (HOPE), and normothermic machine perfusion (NMP) continuously inject the perfusion fluid into the graft blood vessels to form a circuit, mitigating IRI and maintaining cellular metabolism in graft.72 So far, a series of current and ongoing clinical trials have shown that they were superior in reducing ischemic complications compared with static cold storage (SCS).73–76 In addition, the development of detection technology and MP have facilitated the discovery of a series of novel perfusate biomarkers for graft viability evaluation and are summarized as below and in Table 2.5,77–83 Bile production and bile composition (e.g., bile glucose and Na+) during NMP are useful biomarkers for graft synthesis function. Pavel et al.77 restored five discarded DCD livers with NMP for 12 h and found that earlier production of bile and higher bile flows during NMP contributed to better bile duct histology. In addition, Matton et al.78 showed that high biliary bicarbonate and pH, and low biliary glucose in human liver grafts (n=23) during NMP were significantly associated with high risk of bile duct injury. In a porcine LT model, Linares-Cervantes et al.5 demonstrated that a bile/perfusate glucose ratio ≤0.7 and a bile/perfusate Na+ ratio ≥1.1 within 4 h of NMP predicted graft survival after LT. Given that the role of donor graft miRNAs in predicting post-transplant outcomes, perfusate miRNAs may serve similarly. Furthermore, miRNAs have been shown to be stable in perfusate for at least 1 day.79 Verhoeven et al.79 showed that cholangiocyte-derived miRNAs (CDmiRs) in perfusate were predictive of bile duct injury and the development of ITBL. They also found that a significantly elevated hepatocyte-derived miRNA to CDmiRs ratio was associated with the incidence of ITBL. Moreover, Selten et al.80 reported that both high miR-122 levels and a high miR-122/miR-222 ratio in SCS perfusate predicted the development of EAD and poor graft survival after LT in 83 recipients. Flavin mononucleotide (FMN), a critical molecular of generating electrons for ubiquinone reduction in mitochondrial complex 1, was shown to be associated with mitochondrial injury.81 Muller et al.81 preserved 53 grafts with HOPE and demonstrated that a high perfusate FMN level after 30 m of HOPE was strongly linked to severe graft dysfunction. Wang et al.84 infused 23 DCD livers with normothermic regional perfusion and found that the levels of perfusate FMN in transplantable grafts (n=15) were dramatically lower than those in nontransplantable grafts (n=8). D-dimer, a product of fibrin degradation, is a small protein fragment released during fibrinolysis. Karangwa et al.82 preserved 12 discard donor livers with NMP and showed that D-dimer levels >3,500 ng/mL were significantly associated with graft liver injury, suggesting that it was predictive of poor graft function. In a multicenter cohort study, Verhelst et al.83 compared the glycome patterns in SCS perfusate in PNF (n=3) and non-PNF (n=63) groups and found that increased NGA2F, a single under galactosylated biantennary glycan, predicted the development of PNF with 100% accuracy. That highlighted the essential role of omics, especially the metabolomics, in discovering potential perfusate markers of poor graft function during LT.
In recent years, in vivo and ex vivo potential protective interventions that have been used to restore graft function are listed in Table 3.85–102 During the process of ex vivo therapies, the role of MP is apparent because it provides a platform for graft preconditioning.
Previous in vivo studies were performed to treat liver IRI by using small interfering RNA (siRNA). Jiang et al.85 silenced toll-like receptor 4, a critical mediator of inflammation, in a hepatic IRI mouse model, resulting is significant reduction of serum transferases and histological injury. In another study, Zhao et al.86 downregulated nuclear high-mobility group box 1 by transfecting mice with siRNA and found that it effectively inhibited the expression of serum inflammatory cytokines and protected the liver against IRI. Although the efficacy of hydrodynamic injection has been shown in these animal models, it is difficult to use in the clinic because of off-target effects. Recent studies of graft perfusates showed a potential to solve this problem. For example, Gillooly et al.87 found that Fas siRNA directly added to the perfusate was successfully delivered to rat livers during HMP and NMP. This technology ensured that the siRNA only targeted the grafts, opening a new door for graft reconditioning. Antisense oligonucleotide, another gene modulation agent, was demonstrated to significantly reduce miR-122 expression and inhibit hepatitis C virus replication or reinfection after LT in a porcine LT model with NMP, further confirming the possibility of ex vivo gene therapy in grafts.88
In vivo cell therapies such as tolerogenic DCs, Tregs, and mesenchymal stem cells (MSCs) have a role in immunomodulation. In a rat LT model, we innovatively treated acute rejection with a combination of galectin-1-induced tolerogenic DCs and apoptotic lymphocytes, which resulted in prolonged survival of the treated rats, with 37.5% surviving over 100 days, compared with untreated, all of which died within 14 days.90 In a phase I clinical trial, Sanchez-Fueyo et al.90 demonstrated that autologous Tregs transfer was safe and effective in reducing antidonor T cell responses after LT by intravenously administering autologous Tregs to the LT candidates. In addition, Shi et al.91 found that human MSCs injection in LT recipients suppressed acute rejection and improved graft histology by upregulating the Treg/T help 17 cell ratio. Compared with in vivo treatment, ex vivo technology provides novel strategies for graft restoration. For instance, Verstegen et al.92 showed in a porcine LT model that MSCs directly added to the perfusate during HOPE were effectively distributed to the porcine grafts, which continued to maintain their paracrine activity after distribution.
It has been reported that the above tolerogenic cells had the potential to undergo spontaneous malignant transformation.103 Therefore, some investigators began to use MSC-, DC- and trig-derived extracellular vesicles (EVs) as alternatives to cell therapy. In in vivo mice and rat IRI models, MSC-derived EVs had a diverse set of functions including mitochondrial autophagy,104,105 inhibition of immune response106,107 and liver regeneration.108,109 Zheng et al.93 found in a rat IRI model that DC-derived EVs could protect liver against IRI through modulating differentiation of Tregs. In a rat LT model, Chen et al.94 demonstrated that injection with Tregs-derived EVs after LT suppressed the proliferation of CD8+ cytotoxic T cells and prolonged liver graft survival. Compared to the in vivo injection, the ex vivo technology has the potential to directly target donor grafts without concern for off-target effect. Rigo et al.95 successfully delivered human liver stem cells-derived EVs into the rat livers during NMP, leading to less histological damage and lower levels of AST and lactate dehydrogenase in the treated group.
Liver IRI is characterized by the activation of pro-inflammatory responses. Therefore, adding anti-inflammatory agents to perfusate may regulate immune response and alleviate graft damage. In a porcine LT model, Goldaracena et al.96 put alprostadil, n-acetylcysteine, carbon monoxide, and sevoflurane into the NMP perfusate, showing significantly decreased interleukin-6, tumor necrosis factor-α, and AST during NMP, and lower AST and bilirubin levels in serum after LT in the treated group.96 In addition, Yu et al.97 used Mcc950, which strongly inhibited the nucleotide-binding domain leucine-rich repeat containing family pyrin domain containing 3 inflammasome, as an addition to the HMP perfusate in a porcine LT model. They found that Mcc950 significantly reduced inflammatory cytokines and histological injury, and prolonged long-term survival after LT.
During the ischemic phase of LT, rapid adenosine triphosphate depletion and lack of blood flow result in mitochondrial dysfunction and liver sinusoidal endothelial cell (LSEC) injury.110 After reperfusion, the injured LSECs not only produce insufficient vasodilators but also expressed P-selectin to accumulate platelets, which resulted in microcirculation disorder.110 In a rat LT model, Hara et al.98 inhibited the accumulation of platelets by adding prostaglandin E1 (PGE1) to the perfusate under normothermic conditions. PGE1 ameliorated serum liver enzymes and histologic necrosis, and significantly improved bile production and energy status. In addition, Nassar et al.99 added a prostacyclin analog (epoprostenol) to NMP perfusate to preserve porcine livers and found that the use of prostacyclin analog led to high bile production and good histopathology. Furthermore, Echeverri et al.100 compared the effects of endothelin1 antagonist (BQ123), prostacyclin analog (epoprostenol) and calcium channel antagonist (verapamil) to treat hepatic artery vasospasm induced by IRI in a porcine LT model. They demonstrated that grafts with BQ123 and verapamil treatment had higher hepatic artery flow and less hepatocyte injury compared with those treated with epoprostenol.
Moderate to severe (>30%) macrosteatosis is a well-known risk factor for poor graft quality, making it necessary to defat prior to LT.14 Nagrath et al.101 treated rat fatty livers with a combination of six defatting agents normothermically and showed that the treatment could decrease the intracellular lipid content of rat liver by 50% after 3 h perfusion. Furthermore, Boteon et al.102 assessed the efficacy of the above six agents combined with additional L-carnitine in defatting human livers with severe steatosis. They found that this method reduced liver triglycerides and macrosteatosis by 38% and 40% over 6 h NMP, enhanced metabolic parameters including increased urea and bile production, and downregulated biomarkers of liver injury (e.g., lower ALT and reduced inflammatory cytokines).
In addition to the above agents, human atrial natriuretic peptide (hANP), heavy water, marine worm super hemoglobin (M101), glycine, relaxin, and polyethylene glycols have been found to alleviate liver injury.111–116 Nigmet et al.111 added hANP, a protective cardiovascular hormone for vascular endothelia, to SCS perfusate to preserve rat livers, showing that hANP supplementation decreased transaminase release, increased bile production, and protected sinusoidal architecture. In a porcine LT model, Alix et al.113 added M101 to SCS perfusate and demonstrated that M101 significantly reduced blood levels of ALT, AST, and tumor necrosis factor α in recipients 3 days following LT. Moreover, Gassner et al.114 used glycine, a simple amino acid that protected sinusoidal cells and hepatocytes, as an addition to NMP rat liver perfusate. They found less sinusoidal dilatation and tissue damage in the treated group.
This review summarized and updated biomarkers in donor blood, liver tissue or graft perfusate to evaluate early graft injury (e.g., EAD, and PNF) and ITBL, and to identify potential therapies for graft repairment during the era of ECD. We focused on studies using human liver grafts and investigations of potential biomarkers involved in anti- or pro-inflammatory processes, which in turn shape immunity, regulate graft IRI, and further influence the development of EAD, PNF, or ITBL following LT. Given that relevant mechanisms of some molecules are lacking, further prospective studies and experiments are urgently needed to clearly understand their roles. Although various biomarkers with available prognostic and diagnostic value in graft quality assessment have been widely explored, few are currently used in clinical practice. Current challenges associated with biomarker discovery research are as follows. Firstly, the sample sizes of these studies were small and mainly limited to single centers, suggesting that large multicenter cohorts or prospective randomized clinical trials are greatly necessary. Another problem is that the studies lack standardized endpoints and control groups.117 Graft quality is commonly considered to be associated with early graft dysfunction or ITBL, yet other complications after LT (e.g., ACR, metabolic disorders, and graft steatosis or fibrosis) are still a matter of substantial debate. Therefore, we primarily summarized biomarkers predictive of EAD, PNF, and ITBL. Current studies mainly focus on finding biomarkers related to early graft injury, do not have prolonged follow-up and overlook long-term complications like ITBL. Importantly, the measurement of biomarkers should be rapid and easy and have high predictive specificity and sensitivity for graft quality. However, detection of potential biomarkers is costly and time consuming. Moreover, biomarkers need to be stable and measurable during graft procurement, preservation, and implantation. Despite the availability of liver biopsies for histological assessment and quantification of liver injury during LT, they are invasive and only represent specific parts of the grafts. On the contrary, perfusates can be collected in large volumes and contain markers from the whole graft. In recent years, MP has constantly advanced, and it use in evaluation of graft viability has gradually increased. Nevertheless, different regions or centers have their own standards to determine graft quality.78,118 More clear international guidelines that could guide the decision for organ usage, discard, or restoration prior to LT are recommended. In addition, we believe that MP could provide a platform for graft preconditioning, making it convenient to explore novel strategies for graft repair. Although high cost and the technical complexity limit wide usage of MP at its current stage, recently completed and ongoing clinical trials will make it an indispensable part of LT.72,73
Click here for additional data file. Click here for additional data file. |
|
PMC9647110 | Jie Ji,Liwei Wu,Jue Wei,Jianye Wu,Chuanyong Guo | The Gut Microbiome and Ferroptosis in MAFLD | 14-07-2022 | Gut microbiota,Ferroptosis,Nonalcoholic fatty liver disease,Gut-liver axis | Metabolic-associated fatty liver disease (MAFLD) is a new disease definition, and is proposed to replace the previous name, nonalcoholic fatty liver disease (NAFLD). Globally, MAFLD/NAFLD is the most common liver disease, with an incidence rate ranging from 6% to 35% in adult populations. The pathogenesis of MAFLD/NAFLD is closely related to insulin resistance (IR), and the genetic susceptibility to acquired metabolic stress-associated liver injury. Similarly, the gut microbiota in MAFLD/NAFLD is being revaluated by scientists, as the gut and liver influence each other via the gut-liver axis. Ferroptosis is a novel form of programmed cell death caused by iron-dependent lipid peroxidation. Emerging evidence suggests that ferroptosis has a key role in the pathological progression of MAFLD/NAFLD, and inhibition of ferroptosis may become a novel therapeutic strategy for the treatment of NAFLD. This review focuses on the main mechanisms behind the promotion of MAFLD/NAFLD occurrence and development by the intestinal microbiota and ferroptosis. It outlines new strategies to target the intestinal microbiota and ferroptosis to facilitate future MAFLD/NAFLD therapies. | The Gut Microbiome and Ferroptosis in MAFLD
Metabolic-associated fatty liver disease (MAFLD) is a new disease definition, and is proposed to replace the previous name, nonalcoholic fatty liver disease (NAFLD). Globally, MAFLD/NAFLD is the most common liver disease, with an incidence rate ranging from 6% to 35% in adult populations. The pathogenesis of MAFLD/NAFLD is closely related to insulin resistance (IR), and the genetic susceptibility to acquired metabolic stress-associated liver injury. Similarly, the gut microbiota in MAFLD/NAFLD is being revaluated by scientists, as the gut and liver influence each other via the gut-liver axis. Ferroptosis is a novel form of programmed cell death caused by iron-dependent lipid peroxidation. Emerging evidence suggests that ferroptosis has a key role in the pathological progression of MAFLD/NAFLD, and inhibition of ferroptosis may become a novel therapeutic strategy for the treatment of NAFLD. This review focuses on the main mechanisms behind the promotion of MAFLD/NAFLD occurrence and development by the intestinal microbiota and ferroptosis. It outlines new strategies to target the intestinal microbiota and ferroptosis to facilitate future MAFLD/NAFLD therapies.
Because of its close association with metabolic diseases and the many challenges faced by previous diagnostic strategies of exclusion, a new disease nomenclature, metabolic-associated fatty liver disease (MAFLD), has been proposed to replace the previous name, nonalcoholic fatty liver disease (NAFLD).1 Globally, MAFLD/NAFLD is the most common liver disease, with an incidence rate between 6% and 35% in adult populations.2 Studies have shown that the long-term existence of NAFL and NASH are important causes of liver cirrhosis and hepatocellular carcinoma (HCC).3 Indeed, the predominant HCC etiology in the USA is MAFLD/NAFLD. NASH is the second most frequent reason for liver transplantation in the USA, and is likely to supersede hepatitis C as the most common cause of transplantation in the future.4 Although MAFLD/NAFLD is not inherently serious, its complications are, and include liver cirrhosis and HCC which seriously affect quality of life or even endanger patient lives. MAFLD/NAFLD occurrence is an extremely complex pathological process that involves a variety of hepatic cells and multiple extrahepatic signals.5 In recent years, immunoinflammatory responses, genetic metabolism, insulin resistance (IR), ferroptosis, and the gut microbiome have been closely associated with MAFLD/NAFLD.2,5 Bacteria, viruses, fungi, and archaea collectively colonize the human intestines, and are known as the gut microbiome. More than 1×1014 microorganisms are found in healthy individuals and comprise more than nine million genes, which is approximately 150 times larger than the human genome.6 Although the human gut microbiome is closely related to host physiological activities, its importance to human health and disease has long been neglected because of inadequate research methods. However, in recent years, technical advances in DNA/RNA sequencing, bioinformatics data analysis, and culture-based microbiology have increased our understanding of microbes in health and disease.7,8 Simultaneously, the increased gut microbiome literature has been instrumental in delineating metabolic diseases, including NAFLD, obesity, cardiovascular disease, carcinoma, and type 2 diabetes mellitus.9,10 Thus, rather than existing as individual pathogens, microbes exist as complex consortia with myriad interactions with their hosts. The liver and intestinal tract are anatomically and functionally related, having both developed from the same germ layer in the embryo.11 Since the gut-liver axis was first proposed by Marshall in 1998, it has attracted much interest in the relationships between liver disease and the intestinal tract.12 The portal vein connects the gut to the liver and provides 70% of its blood supply. The unique anatomical structure of the liver increases its susceptibility to gut bacteria, bacterial products, endotoxins, and microbiome inflammatory molecules.13 Under normal physiological conditions, the intestinal mucosal barrier is the first bodily defense against external pathogen invasion.14 The liver also produces specific antibodies and inflammatory factors that monitor the intestinal mucosa.15 However, under some pathological conditions, these defense mechanisms become disrupted, thereby facilitating bacterial migration outside the gut. In patients with NAFLD, intestinal bacteria migrate through the portal vein into the liver and cause abnormal activation of the immune system, leading to inflammation responses and injury.16 In addition, interactions between the intestine and liver are bidirectional, and hepatogenic inflammatory cytokines thus impair intestinal mucosal barrier function, disrupting tight junctions of the intestinal epithelium, and forming a malignant liver-gut cycle during NAFLD.17,18 Ferroptosis is a novel form of cell death characterized by iron overload and reactive oxygen species (ROS)-dependent accumulation of lipid peroxides. Ferroptosis, morphologically manifests as mitochondrial shrinkage, reduction or disappearance of mitochondrial cristae, and increased mitochondrial membrane density. Like other cell death modes, ferroptosis is tightly regulated by a variety of intracellular metabolic processes, including glutathione (GSH) synthesis, lipid peroxidation, cysteine transport, iron homeostasis, and NADPH.19 In recent years, many studies have found that ferroptosis is involved in the progression of NAFLD, and preliminarily confirmed that ferroptosis of hepatocytes and intrahepatic macrophages can trigger NASH.20 Inhibition of ferroptosis may become a new therapeutic strategy for NAFLD in the future. In this review, we focus on how the gut microbiota and ferroptosis promote NAFLD development via the gut-liver axis and explore gut microbiome potential as a novel diagnostic biomarker and therapeutic strategy for NAFLD.
The liver and intestinal tract are physiologically bidirectional organs. In one direction, the liver excretes bile and other bioactive mediators into the intestinal cavity via the bile duct, while in the other direction, metabolic nutrients are transported into the liver via the portal vein after reabsorption from the small intestine.11 Simultaneously, intestinal bacteria and their products, e.g., vitamins, short-chain fatty acids (SCFAs), lipopolysaccharide (LPS), endogenous ethanol, and other metabolites are transported through the portal vein, exposing the liver to intestinal microenvironments and pathological changes.17,21
BAs are small molecules synthesized from cholesterol via cholesterol 7a-hydroxylase (CYP7A1) catalysis by liver cells.22 They not only participate in lipid digestion and absorption, but are also important signal regulators that affect energy metabolism, inflammation, and development of liver disease.23 Recent studies have reported that interactions between BAs and intestinal microbiota are closely related to NAFLD.24 BA synthesis is highly complex and includes multistep reactions involving at least 17 different catalytic enzymes (Fig. 1).25 Synthesis occurs in the liver and is accomplished via two different steps. Under normal physiological conditions, at least 75% of BA is synthesized by the classical pathway, which is initiated by cholesterol 7a-hydroxylation catalyzed by CYP7A1.26 CYP7A1 is the rate-limiting enzyme in the process and determines total BAs production.27 The selective pathway is initiated by sterol-27-hydroxylase (CYP27A1) and is further hydroxylated by hydroxysterol 7a-hydroxylase (CYP7B1).28 Studies have shown that the gut microbiota regulates the expression of key enzymes in BA synthesis, including CYP7A1, CYP7B1, and CYP27A1.29 Sayin et al.30 confirmed that liver-based CYP7A1 was regulated by gut microbiota via farnesoid X receptor (FXR)-dependent mechanisms throughout the enterohepatic system in germ-free and conventionally raised mice.30 Moreover, recent research confirmed that inhibiting the intestinal microbiota of hamsters up-regulated CYP7B1 in the alternative BAs synthesis pathway, increased BAs hydrophilicity, and increased tauro-β-muricholic acid (TβMCA).31
The intestinal barrier is an important bodily defense mechanism and is composed of mechanical, chemical, biological, and immune barriers.14 Under normal physiological conditions, large numbers of anaerobic bacteria grow in the intestinal lumen or intestinal mucosa surfaces and include Bifidobacterium that adhere closely to the intestinal epithelium to form a membrane barrier that resists and repels invasion by foreign pathogens.32 Studies have shown that the intestinal microbiota maintain intestinal barrier stability by producing a series of metabolites and instigating signal pathways. Ijssennagger et al.33 reported that sulfide produced by sulfate-reducing bacteria dissolved the mucin polymer network, thinned the mucus layer, and changed the mechanical barrier of the intestinal mucosa. In addition, the Bacteriodes fragilis toxin had proteolytic enzyme-like activity that degraded mucin and destroyed mucus layer structures.34 Furthermore, SCFAs like acetic, propionic, and butyric acids, which are the main metabolites of colonic bacteria required for carbohydrate fermentation, protect the chemical barrier of the intestinal mucosa. Researchers transplanted the butyric acid-producing bacteria, Butyrivibrio fibrisolvens into sterile mice and observed that bacteria restored energy metabolism to colonic epithelial cells and restored cell oxidative phosphorylation and ATP levels, maintained energy homeostasis, inhibited autophagy, and protected colonic epithelial cell integrity.35 More important, the intestinal microbiota are important elements of the intestinal biological barrier; their mechanism of action toward intestinal barrier function is to primarily secrete bacteriocins to kill pathogenic bacteria,36 antagonize pathogen colonization,37 and compete for oxygen and nutrients.38 Destruction of one or more of the barriers affects intestinal barrier integrity. The main driving factors for increased permeability are intestinal inflammation and dysbiosis,39 which are related to long-term antibiotic use,40 chronic alcohol intake,41 continuous high-fat diets,42 and immune-mediated inflammatory disease.43 Akkermansia muciniphila is a Gram-negative anaerobic bacterium that colonizes intestinal mucus layers and is an important link between the intestinal microbiota, inflammation, and intestinal barrier integrity.44 Decreased abundance of A. muciniphila is related to thinning of the mucus layer and increased inflammation, which promotes alcoholic and nonalcoholic liver damage.45 When intestinal permeability increases, microorganisms and microorganism-derived molecules are transferred to the liver through the gut-liver axis causing inflammation and liver damage.46 Some translocated intestinal metabolites may directly interact with host factors, leading to liver disease.18,47 The next section discusses the influence of the gut microbiota on NAFLD and underlying mechanisms.
During embryological development, the gut and liver are intrinsically connected, with the liver budding directly from the foregut during this period. Increasing evidence shows that the intestine and liver have multiple interdependence levels and that dysbiosis and metabolic changes in intestinal microbiota are closely associated with NAFLD (Table 1).45,48–53 This includes observations that patients with NAFLD experience increased intestinal permeability when compared with non-NAFLD patients,54 exhibit correlations between liver disease and microbiota changes,55 and the impact of flora manipulation on liver injury.16
Dysbiosis refers to the destruction of the normal intestinal microbiota, including the loss of beneficial bacteria, changes in bacterial abundance, and increased pathogen levels.6 The condition is induced by factors that include drastic environmental changes, immune or host factors, changes in bile composition, gastric pH, and intestinal motility disorders.6,56 In recent years, studies linking dysbiosis with NAFLD pathogenesis have rapidly increased, focusing on the metabolism of intestinal microbes and their metabolites. However, the exact mechanism by which the gut microbiota promotes the progression of NAFLD needs additional study, and it is also necessary to discover more effective new treatments for gut microbes in NAFLD. A 2001 study by Wigg et al.57 was the first to describe the link between gut dysbiosis and liver disease. Using a 14C-D-xylose-lactulose breath test, the study showed that small intestinal bacterial overgrowth (SIBO) was present in 50% of patients with nonalcoholic steatosis, but in only 22% of control subjects (p=0.048). However, low participant numbers and excluded diseases potentially affected SIBO, such as diabetes and anemia, were major study limitations. In addition, subsequent studies showed that SIBO was related to low intestinal motility and other factors such as the inhibition of gastric acid secretion, decreased secretion of intestinal enzymes, and decreased bile flow, which is a causative factor in NAFLD.58,59 Furthermore, patients with SIBO experienced increased intestinal permeability with more severe portal endotoxemia that may have exacerbated NAFLD progression.48 Animal studies where the microbiome is manipulated provide powerful evidence of dysbiosis in NAFLD. Turnbaugh et al.60 reported that obesity was related to changes in the relative abundance of two main bacteria, Bacteroides and Firmicutes by comparing the gut microbiota of genetically ob/ob mice with lean littermates. They also showed that the ability that the obese microbiota to obtain energy from the diet was partially transmissible, for a significant increase in total body fat after colonizing obese flora in sterile mice compared with the lean flora group. Furthermore, transgenic mouse models have been used to study NAFLD-related intestinal dysbiosis to unravel mechanisms underpinning liver disease progression. Rahman et al.61 used F11r (−/−) mice encoding junctional adhesion molecule A (JAM-A) found that JAM-A deficiency led to more severe NASH. Associated inflammation was reduced by antibiotics, which emphasized the contribution of microbial dysbiosis to NASH development. Although some animal studies have emphasized the role of gut microbiota in NAFLD, the literature on intestinal dysbiosis in human NAFLD is scarce, especially on the full spectrum of NAFLD lesions. An obesity study reported that an increased abundance of Bacteroides was related to percentage weight loss but not to changes in dietary calorie levels. The study was performed by sequencing 16S ribosomal RNA genes from stool samples, suggesting obesity displayed correlations with gut microbiota changes.49 Another study using 16S ribosomal RNA gene sequencing in 57 patients with biopsy-proven NAFLD, revealed that Bacteroides and Ruminococcus were significantly increased, whereas Prevotella abundance was decreased in those with NASH compared with those without the condition.50 Indeed, studies of fecal microbiota transplantation have provided direct evidence. In one study, obese patients with metabolic syndrome received small intestinal infusions of allogenic microbiota from a thin male donor with a body mass index (BMI) <23 kg/m2 or autologous microbiota. Six weeks after infusion, recipient insulin sensitivity and intestinal butyrate-producing microbiota levels were both significantly increased.62 The findings suggest that gut microbiota changes may be used to improve human insulin sensitivity, indicating the potential benefit for NAFLD treatment.
As the hepatic portal vein collects blood supplies from the intestine, the liver is often exposed to potentially harmful intestinal metabolites, including translocated bacteria, LPS, endotoxins, and secreted cytokines.63 Therefore, leaky gut, previously associated with liver disease, has attracted considerable attention in recent decades, and has been widely associated with complementary/alternative medicine approaches.64 Leaky gut is typically caused by several pathogenic factors, including high-fat diet, gut microbiota dysbiosis, and reduced BAs secretion. The conditions change the intestinal mucosal barrier, which increases intestinal mucosa permeability, causing leakage of bacteria, toxic digestive metabolites, and bacterial toxins into the blood, inducing liver immune responses.6,63,65 Dysbiosis changes tight junction proteins in the intestinal mucosa, increases mucosa permeability, and exposes intestinal mucosal cells and the liver to potentially pro-inflammatory bacterial products. Cani et al.66 reported that gut dysbiosis induced by obesity increased lower plasma LPS and cytokine levels and increased the expression of inflammatory and oxidative stress markers associated with higher intestinal permeability and tight junction integrity changes. Meanwhile, gut microbiota are reported to have positive effects on intestinal barriers and permeability. For example, Bifidobacteria was shown to enhance barrier function in experimental necrotizing enterocolitis in mice and the yeast Saccharomyces boulardii had beneficial effects on altered intestinal microbiota and epithelial barrier defects in different pathologies.67 Products from translocated microorganisms may participate in NAFLD pathogenesis through a variety of mechanisms. LPS is the central component of the outer membrane of Gram-negative bacteria and is an endotoxin related to NAFLD pathogenesis. Studies have shown that plasma LPS-binding proteins in patients with NAFLD are significantly increased.68 LPS binds to LPS-binding proteins than then bind to toll-like receptor 4 (TLR4), triggering IR and inflammation.69 During NAFLD occurrence and development, gut dysbiosis leads to increased LPS secretion. SIBO, changed intestinal barrier, and increased permeability promotes circulating LPS level, which then elevated portal levels of gut-derived TLR ligands. Activated TLR4 on hepatic Kupffer cells and stellate cells further stimulated pro-inflammatory and profibrotic pathways via a range of cytokines, including interleukin-1 (IL1), IL6, and tumor necrosis factor (TNF).56,70,71 TLR signal proteins have complex and cooperative interactions with inflammasomes in metabolic diseases.72 Henao-Mejia et al.72 reported that inflammasome-deficient mice had an increased expression of TLR4 and TLR9 agonists and more severe liver steatosis, which were closely related to an imbalance of intestinal microbiota. In fact, TLR signaling enhanced NASH progression by increasing the expression of pro-inflammatory cytokines, such as TNF-α. Specifically, TNF-α regulates liver cell death and prevents insulin signal transduction by inhibiting the insulin receptor and insulin receptor substrate-1, leading to IR.73 Inflammasomes have also been shown to activate several liver processes, including cleavage of pro-caspase 1 to active caspase 1 leading to cell apoptosis.74 Another downstream effect mediated by inflammasomes is the release of IL1β, which promotes NAFLD progression. IL1β regulates lipid metabolism by inhibiting peroxisome proliferator-activated receptor alpha (PPARα) and downstream molecules, leading to accumulation of triglycerides in the liver and promoting steatosis.73
Studies that evaluated the metabolic characteristics associated with NAFLD or NAFLD-fibrosis and are summarized elsewhere.75 Changes in metabolites, including molecules produced by intestinal microorganisms, e.g., ethanol,76 SCFAs such as butyric, propionic, and acetic acid,56 and BA metabolites that target FXR in the liver or intestine,17,77,78 all have important roles in liver injury pathophysiology. Here, we discuss the role of intestinal microbial metabolic substrates and circulating intestinal microbial-derived metabolites in promoting NAFLD progression.
BAs are synthesized by hepatocytes and are discharged into the intestinal tract via the large papilla of the duodenum. Their physiological functions include promoting fat digestion, increasing pancreatic lipase and lipoprotein esterase activity, and regulating the intestinal microbiota.79 BA metabolism (enterohepatic circulation) and associated interactions with gut microbes are extremely complex and have been discussed earlier. In recent decades, BA functions in the pathogenesis and treatment of the fatty liver have received considerable attention and are discussed in several reviews.29,80,81 As a signal regulator molecules, BAs regulate bodily immune homeostasis and inflammatory responses via the FXR (also known as NR1H4) and the G protein-coupled BA receptor, Gpbar1 (TGR-5; also known as GPR131, GPBAR1, M-BAR, and BG37), and further affect the physiological processes of liver cell fatty degeneration, cell damage, and apoptosis.82 FXR is a nuclear receptor believed to be the master regulator of BA metabolism. It is involved in all phases of the biosynthetic pathway and is expressed in a variety of tissues and organs, with the highest expression in liver and ileum cells.83 In addition, the FXR is activated by BAs to inhibit NLRP3 inflammasome activation by interacting with caspase-1, and to reduce release of IL-1B and other inflammatory factors to relieve NAFLD.84 A recent study reported that FXR knockout mice had an increased proportion of secondary BAs and infiltration of lymphocytes and neutrophils, whereas FXR overexpression alleviated liver damage caused by inflammation and infection.85 Indeed, FXR signaling is modulated by the gut microbiota. Li et al.86 used the antioxidant, Tempol, to alter the microbiota and BA distribution, resulting in increased TβMCA levels and suppressed FXR signaling. TGR-5 is another BA response receptor involved in host metabolism. Functioning as a plasma membrane-bound G protein-coupled receptor (GPCR), the protein is generally highly expressed in the gallbladder, placenta, lung, spleen, intestine, liver, brown and white adipose tissue, skeletal muscle, and bone marrow.87 Recently, TGR-5 was shown to have a key role in anti-inflammatory effects,88 reinforcing barrier functions,89 and regulating BA metabolism in participation with intestinal microbiota.23,90 TGR-5 knockout mice had conventional phenotypes and reproductive abilities, but their BA pool was significantly reduced, suggesting that the TGR-5 receptor had important roles in maintaining BA homeostasis.91 In addition, it was demonstrated that treating obese db/db mice with INT-767, a TGR-5 agonist, reduced liver steatosis and inhibited the expression of pro-inflammatory cytokines, indicating the TGR-5 signaling pathway had the potential to treat NAFLD.77
SCFAs are organic fatty acids with 1–6 carbon atoms that are produced by microbial carbohydrate fermentation in the intestinal tract. The most common SCFAs are acetic acid, produced by both the host and bacteria; propionic acid, butyric acid, produced by bacterial fermentation; isovaleric acid, and valeric acid. Acetic, propionic, and butyric acid account for more than 95% of the entire SCFA complement.92 Butyrate is an energy source for intestinal cells and helps maintain the intestinal barrier.6 Recently, it was shown that SCFAs inhibited cell proliferation,93 induced cell differentiation and apoptosis,94 and is closely associated with inflammatory bowel disease (IBD),95 irritable bowel syndrome (IBS),96 colon cancer,97 NAFLD,56,98 and other digestive diseases. SCFA types and levels in the intestine vary with carbohydrate consumption and gut dysbiosis, but they promote NAFLD progression via several mechanisms such as binding to GPCRs. Using isotope-labeled SCFA enemas in rats, Besten et al.99 found that acetic acid, propionic acid, and butyric acid were involved in the expression of fat metabolism-related genes. SCFAs protected the liver by reducing intestinal mucosa permeability through the gut-liver axis and inhibiting endotoxin translocation.99 A recent study by Mollica et al.100 reported that butyric acid and its synthetic derivative, N-(1-carbamoyl-2-phenyl-ethyl) butyramide (FBA), regulated mitochondrial function, efficiency, and kinetics, and proposed it as a new therapeutic strategy to combat obesity and IR. Specifically, butyric acid and FBA improved respiratory capacity and fatty acid oxidation, activated the AMPK acetyl-CoA carboxylase pathway, and promoted inefficient metabolism, thereby reducing intracellular lipid accumulation and oxidative stress.100 Moreover, in another study, acetic acid inhibited liver fat accumulation without changing food consumption or skeletal muscle weight, and was also associated with the PPARα and AMPK pathways.101 Notably, an NAFLD study demonstrated statistically significant differences in Clostridium and Bacteroidetes percentages compared with normal groups. Indeed, the changes between Clostridium and Bacteroidetes may adjust the proportion of SCFAs that affect the energy supply and demand in the liver, altering the progress of NAFLD.102 The GPCRs, GPR41 and GPR43 are the main targets of SCFAs acting on intestinal endocrine cells, and produce a variety of effects that may lead to NAFLD. The exact mechanisms are related to the slowing of gastric emptying and intestinal transit and improved nutrient absorption,103 inhibiting lipolysis and promoting fat cell differentiation,16,104 and increasing intestinal inflammation and permeability to participate in NASH pathogenesis.105
Endogenous alcohol refers to ethanol produced by dietary sugar fermentation, with intestinal microbiota being the main source of this alcohol. Under normal physiological conditions, the body’s metabolism will continuously produce ethanol.106 After eating nonalcoholic food, the blood ethanol concentration also increases. Bacterially-derived ethanol is quickly and completely eliminated in the portal vein by liver alcohol dehydrogenase (ADH), catalase, and the microsomal ethanol oxidizing system. When ADH is inhibited, blood ethanol concentrations increase. The fact that the human liver and digestive tract both have the highest ADH activities proves that the intestinal tract produces alcohol.107 NAFLD and alcoholic fatty liver disease are pathologically similar and may have common pathogenic mechanisms. Studies have confirmed that blood ethanol concentrations are higher in obese patients or obese mice than in lean individuals, suggesting intestinal alcohol may be related to the occurrence of NASH.108 In addition, excess growth of small intestinal bacteria and gut dysbiosis (e.g., increased Escherichia coli) may lead to increased levels of endogenous alcohol. Zhu et al.109 reported significantly increased E. coli levels in patients with NASH compared with obese patients. As E. coli is the main alcohol-producing bacteria, differences in blood ethanol concentrations were observed, suggesting a role for alcohol-producing microbiota in this condition. Moreover, recent studies reported the increased expression of alcohol-metabolizing enzymes (i.e. ADH) in patients with NASH. Specifically, increased ADH activity increased acetaldehyde levels, which further increased small intestine mucosa permeability. The absorption of intestinal microbiota metabolites increased, which augmented acetaldehyde levels and promoted NASH.110
Iron overload is prevalent in NAFLD patients, and it is widely accepted that iron-induced lipid peroxidation is one of the major triggers of NAFLD.111 In addition, iron imbalance is associated with obesity and IR, both of which are typical features of patients with NAFLD.112 In general, people tend to speculate that ferroptosis may be involved in the pathogenesis of NAFLD, which has been confirmed by numerous studies.113 Fortunately, some drugs that act on ferroptosis targets (e.g., sorafenib, sulfasalazine, and artesunate) have been widely reported, making it possible that ferroptosis could be a key target for the treatment of NAFLD (Table 2).114–126 Dietary Fe3+ is absorbed by duodenal intestinal epithelial cells and reduced to Fe2+ by divalent metal-ion transporter-1 (DMT1). Fe2+ absorbed into the blood is oxidized to Fe3+ by ceruloplasmin, bound by transferrin, and then transported to tissues. However, because of the first-pass effect of the hepatic portal circulation, iron exposure of the liver is much greater than that of other tissues, resulting in liver damage and various complications.127 Serum ferritin is a clinical biomarker for detecting iron homeostasis in the body. When the serum ferritin content is abnormal, the overload operation of the liver as an organ responsible for removing serum ferritin further aggravates liver damage. The expression of ferritin is influenced by iron stores and inflammation, and elevated ferritin levels are common in NAFLD.128 In a study of 628 adult patients with biopsy-proven NAFLD, a serum ferritin (SF) 1.5 times the upper limit of normal has been associated with a diagnosis of NASH, higher steatosis grade, and lobular inflammation. Elevated SF was also found to be an independent predictor of advanced hepatic fibrosis in patients with NAFLD.129 Also, a study of 25,597 participants in Korean National Health and Nutritional Examination Surveys between 2007 and 2012 and confirmed that people with higher SF levels were more likely to have NAFLD. An increase in SF of 10 ng/mL increased the likelihood of NAFLD by 3–10%.130 Therefore, many researchers have proposed that in equivocal circumstances, SF measurement can be used to assess the risk of NAFLD.131 However more long-term studies are needed to assess the relationship between SF levels and complications of liver disease (e.g., HCC) and liver-related mortality. Iron overload is prevalent in NAFLD patients.132 In a retrospective study, patients with biopsy-proven NAFLD and iron overload had poor long-term outcomes133 that may have been the result of increased IR, excess hepatic lipid peroxidation, and accelerated liver fibrosis progression caused by iron overload.134 Loguercio et al.135 found that more than 90% of NAFLD patients had increased levels of lipid peroxidation markers, including malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE), which were significantly higher in NASH patients than in steatosis patients. Qi et al.113 studies the impact of ferroptosis on the progression of NASH induced by a methionine/choline-deficient diet (MCD) for 10 days. RSL3, a ferroptosis activator, aggravated symptoms, including serum biochemical index levels, liver steatosis, and inflammation) in mice with NASH induced by the MCD diet. Sodium selenite, a GPx4 activator, rescued RSL3-induced lipid peroxidation and cell death. Similarly, Li et al.20 used RNA-seq analysis to show that arachidonic acid metabolism promote ferroptosis in the MCD diet-induced NASH mouse model, suggesting that ferroptosis may be a therapeutic target for NASH treatment. Consistently, other studies found that some drugs like Ginkgolide B and dehydroabietic acid alleviated NASH severity by inhibiting ferroptosis. In that context, Nrf2 and GPx4 stand out as major protective mechanisms.114,136 Overall, the results imply that the regulation of ferroptosis in the context of NAFLD is an intriguing notion that deserves further investigation.
As discussed, gut dysbiosis and associated metabolites such as BAs, SCFAs, and endogenous ethanol, and inflammatory responses and damage of the intestinal barrier are important factors during NAFLD occurrence and development. If those conditions are corrected, NAFLD progression can be slowed and possibly reversed. This section focuses on gut microbiota regulation as a therapeutic target for NAFLD prevention, including lifestyle and diet therapies, antibiotics, probiotics, and prebiotics, glucagon-like peptide-1 (GLP-1) receptor agonists (GLP-1 RA), and sodium/glucose cotransporter-2 inhibitors (SGLT2i; Table 3).[137,–146]
NAFLD is closely related to obesity.147 Studies have shown that eating foods rich in fat and fructose alters the intestinal microbiota, changes intestinal barrier function, and causes endotoxemia and inflammatory reactions, all of which promote obesity and NAFLD.148 Therefore, the most important treatment goal for patients with NAFLD is weight reduction and maintenance of a healthy lifestyle to reduce liver fat deposition and inflammatory responses. In addition, a balanced diet, adequate sleep, and appropriate exercise are essential to maintain intestinal microbiota stability and health, and to reduce the risk of other diseases. Dietary interventions are effective in the treatment of NAFLD patients. Even a modest 3–5 kg weight gain predicts the development of NAFLD independent of baseline BMI. In addition, patients with NAFLD were found to experience a 75% remission rate with a weight loss of 5% or more from baseline.149 Much evidence suggests that the Mediterranean diet can reduce liver fat, even without weight loss. It is the most recommended diet for NAFLD.150 The Mediterranean diet includes nuts, fruits, legumes, olive oil, vegetables, and fish, in which consumption of sugar and refined carbohydrates is decreased and consumption of monounsaturated fatty acids and omega-3 fatty acids is increased. Dietary micronutrients also greatly influence the progression of NAFLD. Studies have shown that the intake of micronutrients such as vitamin C, vitamin D, and choline is significantly negatively correlated with the prevalence of NAFLD, which may be related to their antioxidant and antifibrotic activity.151 However, randomized controlled trials have not resulted in clear evidence that high-dose vitamin D supplementation is beneficial for hepatic steatosis or IR in NAFLD.152 Although the Mediterranean diet advocates moderate alcohol consumption, whether or not alcohol should be allowed in NAFLD patients remains controversial. Regular alcohol consumption increases the risk of developing HCC in NASH patients with cirrhosis, so alcohol should be avoided in such patients.153 However, in patients without cirrhosis, uncertainty about the impact of moderate drinking (e.g., 12 ounces of beer, 4 ounces of wine, or 1 ounce of liquor) is not clear, so prudent drinking is advised. Sleep is an essential physiological process required for normal function, and adequate sleep patterns help maintain normal homeostasis. In recent decades, studies have shown that sleep-related factors, especially sleep time, influence the risk of obesity.154 A prospective study by Nielsen et al.155 reported that patients with obesity slept for significantly shorter periods than nonobese individuals. Also, Gildner et al.156 observed that in middle-aged individuals, short sleep duration was significantly correlated with elevated BMI and waist circumference. However, obesity-associated mechanisms caused by shortened sleep times remain unclear. A recent study showed that a chronic lack of sleep decreased leptin and increased ghrelin levels, resulting in a “hyperappetite.”157 Increased eating rate caused by prolonged wakefulness was also a cause of obesity.158 People with short sleep times are prone to fatigue that leads to reduced exercise, increased weight gain, or obesity. In addition to sleep time, changes in circadian rhythm influence development of obesity and NAFLD progression.137 Voigt et al.159 reported that reversing circadian rhythms in mice changed the Firmicutes and Bacteroidetes composition in those fed a high-sugar diet, but the microbiome in mice fed normal diets did not change. Summa et al.160 also found that circadian rhythm disorders increased intestinal permeability in mice, promoting alcohol-induced steatohepatitis.160 Weight loss is recognized as a basic and key measure for NAFLD management, and exercise is an effective and safe way to lose weight. For patients with NAFLD, exercise not only directly reduced liver fat content, but also reduced fatty acid absorption, improved insulin sensitivity,161 reduced liver transaminase indicators, and improved other metabolic indicators.162 In 2009, George et al.162 compared the low intensity exercise, medium intensity exercise, and control groups. After a 3-month intervention, patients who maintained more than 150 min/week had decreased serum transaminase levels, independent of changes in body mass, effectively illustrating physiological exercise advantages for these patients. Previous studies have also shown that exercising changes the composition of intestinal microbes and affects NAFLD progression.163 Munukka et al.164 recruited 17 overweight adult women for a 6-week bicycle endurance training study and found that exercises decreased Proteobacteria, but significantly increased Akkermansia levels. Animal studies have also confirmed the effect of exercises on gut microbiota. Petriz et al.165 reported that treadmill exercises changed the composition and abundance of microorganisms, and training increased lactobacilli, (beneficial bacteria) numbers in obese rats. Denou et al.138 conducted a 6-week high-intensity exercise regime in rats fed a high-fat diet, and found that the exercise intervention increased gut microbe diversity, improved metabolic capacity, and reduced the Bacteroides to Firmicutes ratio.
Therapeutic antibiotics inhibit excessive proliferation of intestinal microbes and bacterial translocation. They alter disease-related microbial communities to ensure healthy homeostasis. Antibiotics eliminate harmful microbiota, and are effective in several digestive-system models, including hepatic encephalopathy,166 IBS,[1167] IBD,168 and NAFLD.47,169 The therapeutic effects of antibiotics in NAFLD are attributed to (1) improving leaky gut by reducing pathogens and potential pathogens and suppressing liver inflammation and (2) reducing harmful bacterial metabolites which promote NAFLD. In an observational study, Gangarapu et al.139 reported that after 28 days of with rifaximin treatment of NASH patients, circulating endotoxin levels were significantly reduced. Short-term rifaximin treatment acted on intestinal bacteria to achieve the desired therapeutic effectiveness. However, more clinical trials are required to confirm effective rifaximin treatment cycles for patients with NAFLD. Animal studies have reported that antibiotics quickly and significantly altered the intestinal microbiota. Broad-spectrum antibiotics, such as ampicillin, neomycin, metronidazole, and vancomycin reduced hepatitis by regulating free secondary BA levels and improving liver steatosis by inhibiting the FXR pathway to downregulate liver SREBP1C expression.85,170 However, another study found that penicillin G and erythromycin aggravated liver lipid metabolism and inflammatory reactions.171 Antibiotics also promoted liver lipid accumulation, inflammatory responses, and liver fibrosis by increasing liver immune damage.172 At the same time, we cannot ignore the impact of antibiotics on the beneficial intestinal microbiota, therefore correct use of effective antibiotic treatment requires more comprehensive research.
Probiotics are living microorganisms that benefit host health.6 Studies show that they regulate the intestinal microbiota,173 enhance intestinal barrier function,174 reduce intestinal permeability,175 alleviate immune and metabolic damage,176 up-regulate fatty acid oxidation,177 and reduce liver steatosis and inflammatory-response damage.178 A recent meta-analysis confirmed that Lactobacillus, Bifidobacterium, Streptococcus probiotics, when used for 8–24 weeks were beneficial for the recovery of liver enzymes and IR in patients with NAFLD.179 A clinical study reported that after a 6 month intervention with the probiotic, Lepicol in 10 patients with NASH, intrahepatic triacylglycerol levels were reduced by more than 30% compared with baseline levels and serum AST levels were significantly reduced.180 A randomized controlled trial of 42 patients with NAFLD found that fasting blood glucose, insulin, IR, TNF-a, and IL6 were significantly reduced after 8 weeks of probiotic treatment (two capsules/day).181 Other studies have reported an association of probiotics on liver fibrosis or death in NAFLD patients, therefore results are inconsistent. A liver biopsy clinical study reported that Bifidobacterium longum supplementation significantly improved liver steatosis, but not liver fibrosis.182 In a long-term survey of 39 biopsy-confirmed patients with NAFLD, the continuous use of the probiotic, VSL#3 for 1 year significantly improved NAFLD activity scores, hepatocyte swelling, and liver fibrosis. Prebiotics are dietary supplements that benefit the host by selectively stimulating the growth and/or activity of one or several bacterial colonies.183 Matsumoto et al.140 studied the effects of fructo-oligosaccharides (FOSs) on intestinal barrier function and steatohepatitis in mice with methionine-choline deficiency. Liver inflammation and hepatocyte steatosis in FOS-treated mice were significantly reduced (p<0.01), suggesting that 3 weeks of treatment improved NAFLD and restored barrier functions in the intestinal tract.140 The probiotic lactulose promoted Bifidobacteria and lactic acid bacteria growth. Fan et al.141 used it to treat mice with NAFLD induced by a high-fat diet, and showed that liver inflammation indicators such as AST and ALT in the lactulose treatment group (0.9 mL/kg/day for 8 weeks) were significantly reduced, but hepatocyte steatosis was not significantly improved, suggesting that lactulose reduced liver inflammation but did not improve fat degeneration in liver cells.141
GLP-1 is an incretin secreted by L cells in the distal small intestine and colonic mucosa after meal stimulation in a glucose concentration-dependent manner. It promotes insulin secretion and participates in the regulation of blood glucose homeostasis. GLP-1 has a very short half-life in vivo, and is degraded by dipeptidyl peptidase-4 (DPP-4), so it cannot be used for disease treatment.184 GLP-1 RA belongs is an incretin drug with pleiotropic effects such as lowering blood glucose and blood lipids and reducing body weight.185 Recent studies have found that GLP-1 RA improved IR in NAFLD, reduced liver steatosis, and improved liver fibrosis. It is of great significance for the treatment of NAFLD, especially NAFLD complicated with T2DM.186 A randomized, multicenter, double-blind, placebo-controlled phase 2 trial in the UK that evaluated the safety and efficacy of subcutaneous liraglutide, an acylated GLP-1 RA, 1.8 mg daily compared with placebo in patients with biopsy-confirmed NASH. Liraglutide significantly improved hepatic steatosis by 83% in the liraglutide group and 45% in the placebo group, and hepatocyte swelling by 61% in the liraglutide group and 32% in the placebo group. which indicated that the patient’s NASH was in remission. The histological effects of liraglutide on NASH were not entirely mediated by its action on the improvement of glycemic control.187 A study by Moreira et al.142 showed that liraglutide not only reduced hepatic fat accumulation by 78% in ob/ob mice and reversed steatosis in HFD mice, but also altered the overall gut microbial composition. Proteobacteria decreased and Akkermansia muciniphila increased in the mice fed the HFD. The studies suggest that GLP-1 RA contributed to the improvement of NAFLD by the regulation of gut microbiota, which offers a new perspective for us to find gut microbiota-targeted therapies of NAFLD. Sodium-glucose cotransporter 2 (SGLT-2) inhibitors are a class of hypoglycemic drugs that is commonly used in clinical practice to reduce the reabsorption of glucose by the kidneys, intestines, and heart. Several studies showed that SGLT-2i was associated with improvement of hepatic steatosis.143 In an open-label, randomized, active-controlled trial, Ito et al.189 of 66 patients with type 2 diabetes and NAFLD found that ipragliflozin 50 mg significantly reduced body weight and visceral fat area. A similar study by Nasiri-Ansari et al.190 in mice fed an HFD found that empagliflozin reduced fasting glucose, total cholesterol, and serum triglyceride levels; and decreased the NAFLD activity score, expression of lipogenic enzymes, and inflammatory molecules. However, side effects associated with SGLT-2 inhibitors, such as increased risk of urinary and genital infections cannot be ignored. The increased risk may be explained by the fact that persistent diabetes may promote the growth of pathogenic microorganisms. A meta-analysis showed that gliflozins were associated with an increase in genitourinary infections,191 and they have also been reported to increase the risk of malignancy, particularly of the breast or bladder, but no studies have confirmed that possibility.192
Evidence that the gut microbiota has important mechanistic roles in NAFLD occurrence and progression is increasing. Gut microbiota dysbiosis usually reduces beneficial bacteria and SIBO, changes small intestine mucosal barriers, and increases intestinal permeability and microbial metabolites (e.g., LPS and SCFAs). That increases endotoxins and inflammatory factors that enter the liver via the gut-liver axis, inducing immune and inflammatory reactions, and culminating in NAFLD (Fig. 2). No drugs are currently licensed for NAFLD therapy, but diet and exercise have proven to be effective treatments. Because of the relationship between NAFLD and T2DM, many diabetes drugs have achieved positive results in relieving NASH. In addition, experimental drugs targeting intermediate metabolism in NAFLD have also been shown to be beneficial, but adverse effects may limit their use. This review focuses of the gut microbiota and ferroptosis treatments for NAFLD, as well as proposing new treatment strategies. Lifestyle and diet, antibiotics, regulation of ferroptosis, probiotics, and prebiotics, GLP-1 RA, and SGLT2i may become effective and safe treatments to alleviate NAFLD. However, to effectively transform and apply animal model findings to humans, well-designed large clinical trials, spanning multiple disease etiologies and patient characteristics, are required. |
|
PMC9647112 | Teng Liu,Jing Zhao,Jia-Yan Feng,Yi Lu,Jonathan A. Sheps,Ren-Xue Wang,Jun Han,Victor Ling,Jian-She Wang | Neonatal Dubin-Johnson Syndrome and its Differentiation from Biliary Atresia | 11-03-2022 | Jaundice,ABCC2,Cholestasis,MRP2,Acholic stools | Background and Aims The aim was to determine if liver biochemistry indices can be used as biomarkers to help differentiate patients with neonatal Dubin–Johnson syndrome (nDJS) from those with biliary atresia (BA). Methods Patients with genetically-confirmed nDJS or cholangiographically confirmed BA were retrospectively enrolled and randomly assigned to discovery or verification cohorts. Their liver chemistries, measured during the neonatal period, were compared. Predictive values were calculated by receiver operating characteristic curve analysis. Results A cohort of 53 nDJS patients was recruited, of whom 13 presented with acholic stools, and 14 underwent diagnostic cholangiography or needle liver biopsy to differentiate from BA. Thirty-five patients in the cohort, with complete biochemical information measured during the neonatal period, were compared with 133 infants with cholangiographically confirmed BA. Total and direct bilirubin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bile acids, alkaline phosphatase, and gamma-glutamyl transferase were significantly lower in nDJS than in BA. In the discovery cohort, the areas under the curve for ALT and AST were 0.908 and 0.943, respectively. In the validation cohort, 13/15 patients in the nDJS group were classified as nDJS, and 10/53 in the BA control group were positive (p<0.00001) with an ALT biomarker cutoff value of 75 IU/L. Thirteen of 15 patients were classified as nDJS and none were classified positive in the BA group (13/15 vs. 0/53, p<0.00001) with an AST cutoff of 87 IU/L. Conclusions Having assembled and investigated the largest cohort of nDJS patients reported to date, we found that nDJS patients could be distinguished from BA patients using the serum AST level as a biomarker. The finding may be clinically useful to spare cholestatic nDJS patients unnecessary invasive procedures. | Neonatal Dubin-Johnson Syndrome and its Differentiation from Biliary Atresia
The aim was to determine if liver biochemistry indices can be used as biomarkers to help differentiate patients with neonatal Dubin–Johnson syndrome (nDJS) from those with biliary atresia (BA).
Patients with genetically-confirmed nDJS or cholangiographically confirmed BA were retrospectively enrolled and randomly assigned to discovery or verification cohorts. Their liver chemistries, measured during the neonatal period, were compared. Predictive values were calculated by receiver operating characteristic curve analysis.
A cohort of 53 nDJS patients was recruited, of whom 13 presented with acholic stools, and 14 underwent diagnostic cholangiography or needle liver biopsy to differentiate from BA. Thirty-five patients in the cohort, with complete biochemical information measured during the neonatal period, were compared with 133 infants with cholangiographically confirmed BA. Total and direct bilirubin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bile acids, alkaline phosphatase, and gamma-glutamyl transferase were significantly lower in nDJS than in BA. In the discovery cohort, the areas under the curve for ALT and AST were 0.908 and 0.943, respectively. In the validation cohort, 13/15 patients in the nDJS group were classified as nDJS, and 10/53 in the BA control group were positive (p<0.00001) with an ALT biomarker cutoff value of 75 IU/L. Thirteen of 15 patients were classified as nDJS and none were classified positive in the BA group (13/15 vs. 0/53, p<0.00001) with an AST cutoff of 87 IU/L.
Having assembled and investigated the largest cohort of nDJS patients reported to date, we found that nDJS patients could be distinguished from BA patients using the serum AST level as a biomarker. The finding may be clinically useful to spare cholestatic nDJS patients unnecessary invasive procedures.
Hyperbilirubinemia II, or Dubin–Johnson syndrome (DJS; OMIM 237500), was first reported by Dubin and Johnson in 1954.1 It is an autosomal recessive disorder caused by variations in the ATP-binding cassette subfamily C member 2 (ABCC2) gene that result in a decrease in the production or loss of function of multidrug resistance-associated protein 2 (MRP2).2,3 The disorder presents in adolescence and is characterized by a low-grade elevation of conjugated bilirubin in the blood, and histologically with the accumulation of dark, coarsely granular, melanin-like pigment in centrilobular hepatocytes, with no other signs of hepatic injury.4,5 It has been reported that DJS can present with severe cholestasis and hepatomegaly in neonates (nDJS).6–8 However, with only a limited number of reported cases, the clinical symptoms, pathologic signs, and genetic features of nDJS remain poorly described. Biliary atresia (BA) is the leading cause of pediatric liver transplantation worldwide. It presents as an obliterative cholangiopathy with neonatal jaundice and pale stools.9 Prompt diagnosis and the Kasai procedure improve the odds of native liver survival. However, differentiation from other causes of neonatal cholestasis is difficult because of the low specificity of imaging studies, including ultrasonography, hepatobiliary iminodiacetic acid scans, and magnetic resonance cholangiopancreatography, among others. Quite often, liver biopsy or surgical cholangiography is needed to confirm the diagnosis.9 As nDJS patients usually do not require specific treatment, but BA needs early surgical intervention,10 early differentiation is critical. A previous study in Japan by Togawa et al.,11 reported that eight of ten nDJS patients underwent liver biopsy during neonatal cholestasis because they were suspected of having BA. The primary aim of this study is to better define the clinical and pathologic features of nDJS and to determine whether convenient biomarkers can be identified for helping to distinguish nDJS from BA.
This was a single-center retrospective study. The subjects were Chinese children who were admitted to the Children’s Hospital of Fudan University, between 2011 and 2021 for the study of neonatal cholestasis with consent under a protocol (ethical approvals 2015-178 and 2020-402) in accordance with the ethical guidelines of the 1975 Declaration of Helsinki. nDJS was defined as onset of cholestasis before 3 months of age, with confirmed homozygous or compound-heterozygous variants of uncertain significance, or with pathogenic/likely pathogenic ABCC2 variants, according to American College of Medical Genetics /Association for Molecular Pathology guidelines,12 and after exclusion of other causes such as infection, hemolytic jaundice, endocrine disease, bile acid synthesis disorders, and other metabolic conditions. Of 53 genetically-confirmed nDJS patients (Table 1), the 35 patients with complete liver biochemistry data obtained during the neonatal period were selected as the test subjects. A group of 133 age- and sex-matched cholangiography-confirmed BA patients, with available pre-operation liver biochemistry indices, were used as the control group. Their liver chemistry indices were compared and analyzed by receiver operating characteristic (ROC) curves. A total of 80 patients with BA and 20 with nDJS were randomized to the discovery cohort; in addition, 53 patients with BA and 15 with nDJS were included in the validation cohort. The screening criteria for BA included: (1) Onset of clinical manifestations of BA, such as prolonged jaundice with high GGT levels and acholic stools during the neonatal period; (2) A cholangiogram showing that the bile ducts were not patent during the cholangiography; (3) Biopsy specimens evaluated by two pathologists to ascertain the diagnosis of BA; (4) Exclusion of other causes of neonatal cholestasis through appropriate investigations, such as Alagille syndrome, citrin deficiency, progressive familial intrahepatic cholestasis type 3 and other metabolic conditions. Ninety-four age-matched patients with neonatal intrahepatic cholestasis were chosen as cholestasis controls (Supplementary Table 1). They included four patients with bile acid synthesis defects, three with citrin deficiency, one with progressive familial intrahepatic cholestasis type 2, 21 with Alagille syndrome, and 65 with an unknown etiology, i.e. testing for neonatal hemochromatosis, viral hepatitis, and others. failed to identify an etiology.
The pathogenicity of newly discovered missense variants was predicted by in silico tools including the MutationTaster (http://www.mutationtaster.org/), Polymorphism Phenotyping v2 (Polyphen-2, http://genetics.bwh.harvard.edu/pph2/index.shtml), Sorting Intolerant From Tolerant (SIFT, http://sift.jcvi.org), Protein Variation Effect Analyzer (PROVEAN, http://provean.jcvi.org/index.php), Mendelian Clinically Applicable Pathogenicity (http://bejerano.stanford.edu/MCAP/), and Functional Analysis Through Hidden Markov Models (http://fathmm.biocompute.org.uk/). The pathogenicity of noncanonical splicing variants and nonframeshift insertion/deletion variants was predicted by MutationTaster. Default settings were used for all in silico tools.
Liver biopsy specimens were obtained from one patient with neonatal Dubin–Johnson syndrome during evaluation of the cause of cholestasis and from seven during laparoscopic cholangiography because BA was suspected. The specimens were processed routinely, and 4 µm sections were cut from formalin-fixed, paraffin-embedded blocks and stained with hematoxylin and eosin for routine histologic evaluation of collagen, bile, and melanin-like pigment deposits in hepatocytes.
Statistical analysis was performed with SPSS 19.0 (IBM Corp., Armonk, NY, USA) to determine differences between the nDJS patients and control groups. Kolmogorov-Smirnov tests were performed to check if the data were normally distributed. Student’s t-test was performed when the data had a normal distribution. A nonparametric test, the Mann-Whitney U test, was performed when the data did not have a normal distribution. Data were reported as medians and interquartile range, or means ± SD. ROC curve analysis was used to calculate the areas under the curve (AUC) with 95% confidence intervals (CIs).
During the study period, about 7,000 patients were referred to us for investigation of the causes of neonatal cholestasis. Of those 53 (31 men and 22 women) were diagnosed with nDJS (Table 1). Cholestasis subsided or improved during follow-up in all nDJS patients, and none failed to thrive. Of 28 patients with a recorded date of onset of jaundice during the neonatal period, 25 visited a pediatric gastroenterologist because of jaundice shortly after birth (0–10 days after birth, with an average of 3.6±1.9 days. Another three patients presented with jaundice at the first, second, and third month after birth. Peak jaundice was detected at 2–79 (27.9±19.9) days after birth, during the first 3 months. Peak total bilirubin (TB) ranged from 3.5–29.3mg/dL (13.9±6.3mg/dL) and peak direct bilirubin (DB) ranged from 2.3–22.2mg/dL (6.2±4.5mg/dL). Twenty-two patients were followed-up at our center. Jaundice disappeared in seven patients (P7, P13, P20, P21, P22, P32, P41), declined to subclinical levels (TB levels between 1–2mg/dL) in five (P8, P9, P15, P24, P43), and persisted in six (2–4mg/dL; P5, P12, P16, P27, P29, P42) at the last follow-up. Jaundice in patients P3, P17, and P31 disappeared at follow-up but relapsed as subclinical jaundice of unknown cause. P37 experienced a recurrence of subclinical jaundice at 3.2 years of age after a fever lasting for 2 days. The liver and spleen were palpable 2–3 cm below the costal margin or xiphoid in some cases, but were not palpable in others. Fourteen patients had cholangiography or liver biopsy or both, to verify the presence of BA. Thirteen of 30 patients with descriptions of stool color were reported to have acholic stools. All infants diagnosed with DJS were full-term except for patients P15, P28, and P30 (Table 1). Three (P28, P30, P42) presented with transient hypoglycemia, and three (P6, P34, P40) presented with transient prolonged prothrombin time responsive to vitamin K1 injection. One patient (P18) had low muscle tension. The mother of P13 reported itching during pregnancy, and P28 had retarded brain development. The characteristics of BA patients and cholestasis controls are shown in Supplementary Material 1.
Fifty-five ABCC2 variants were identified in 53 nDJS patients (Table 1). c.1177C>T (p.Arg393Trp), c.3825C>G (p.Tyr1275Ter), c.2755T>A (p.Ser919Thr), c.4239_4240dupTC (p.His1414LeufsTer18), c.3928C>T (p.Arg1310Ter) occurred in seventeen, twelve, seven, seven, and four alleles, respectively; c.1939G>T (p.Glu647Ter), c.2302C>T (p.Arg768Trp), c.338T>C (p.Leu113Pro), c.4024T>C (p.Ser1342Pro) were each found in three alleles, while c.2980delA (p.Ile994LeufsTer29), c.298C>T (p.Arg100Ter), c.3196C>T (p.Arg1066Ter), c.4384delG (p.Glu1462ArgfsTer8), c.4384G>A (p.Glu1462Lys), c.4465_4473delinsGGCCCACAG (p.Ile1489_Ile1491delinsGlyProGln) occurred two times each.
Twenty-two of the variations we observed had been previously reported,10,11,13–16 and thirty-three were novel (Table 2). Of the novel variants, six were canonical splicing variants, two were noncanonical splicing variants, five were nonsense variants, thirteen were missense variants, and seven were frameshift variants. No significant correlations were observed between the type of variant and the presence of acholic stools (Supplementary Material 2, Supplementary Table 2).
Total protein and albumin values were within the normal range in DJS patients (data not shown). We evaluated serum TB, DB, total bile acids (TBAs), alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), and gamma-glutamyl transferase (GGT) in 35 patients during neonatal cholestasis between 16 and 93 days of age. In those nDJS patients the median and interquartile range was determined for TB [7.2 (5.3, 9.1) mg/dL], DB [4.5 (3.3, 5.8) mg/dL], ALT [29.7 (16.5, 45) IU/L], AST [33.5 (23.1, 68.8) IU/L], ALP [386 (326, 524) IU/L], GGT [149 (84.5, 216.5) IU/L], and TBA [72.2 (30.6, 109.1) µmol/L]. TB, DB, ALT, AST, GGT, ALP, and TBA levels were all significantly lower in infants with nDJS than in infants with BA. DB, ALT, AST, and ALP levels were also significantly lower in nDJS patients than in cholestasis controls (Fig. 1). No significant correlations were observed between the type of variant and the levels of TB, DB, ALT, and AST (Supplementary Material 2, Supplementary Fig. 1). In addition, TB, DB, ALT, AST, TBA, ALP, and GGT levels had no significant linear correlation with days after birth in any of the patient groups. (Supplementary Fig. 2).
Liver biopsy specimens were obtained from eight patients at 1–3 months of age, six of whom had undergone cholangiography and one with a needle liver biopsy to verify the presence of BA. Only one patient was reported to have a dark red liver during the procedure, at 40 days of age. Histopathologically, the presence of fibrosis, cirrhosis, or melanin-like pigment deposits in hepatocytes was not observed, except fibrosis in P20. However, giant-cell transformation of hepatocytes/hepatocyte ballooning (6/8), steatosis (4/8), and slight intracanalicular and intracytoplasmic cholestasis, i.e. bile plugs (7/8) were observed. The results of five patients with liver biopsies at our center are shown in Figure 2.
Performance of discriminatory features of biomarkers for nDJS diagnosis: 35 nDJS patients and 133 age-matched subjects with BA were enrolled in the study (Supplementary Fig. 3). To investigate the predictive values of discriminatory features (TB, DB, TBA, ALT, AST, GGT, and ALP levels) for nDJS, we calculated the AUC of each feature in the discovery cohort (Fig. 3). The AUC for ALT in nDJS was 0.908 (95% CI: 0.834–0.982; p<0.0001) and that of AST was 0.943 (95% CI: 0.861–1.000; p<0.0001; Fig. 3E). The sensitivity and specificity for ALT were 81.25% and 90%, respectively, at a cutoff value 75 IU/L. The optimal threshold cutoff for AST in nDJS, was 87 IU/L with a sensitivity of 95% and a specificity of 90%. We tested the discriminating power of ALT and AST using a double-blind strategy in a validation cohort of 68 subjects, 15 with DJS and 53 with BA (Table 3). Thirteen of the 15 nDJS were classified as positive and 10 of the 53 BA patients were classified as positive using ALT as the biomarker. With an AST cutoff value of <87 IU/L, 13 of 15 patients were classified as positive in the nDJS group, and 0 of the 53 BA patients were classified as positive.
Accurate diagnosis of DJS in neonates is important to rule out other hepatobiliary disorders such as BA, which is the most common cause of neonatal cholestasis and usually progresses rapidly, with severe liver injury leading to liver transplantation.5,10 nDJS is a relatively mild, autosomal recessive disorder that does not develop into severe liver injury, but manifests with severe cholestasis and acholic stools during the neonatal period, a presentation that sometimes overlaps with BA. The Kasai hepatic portoenterostomy procedure for re-establishing bile flow is the most effective surgical intervention for survival in BA. The liver is preserved, and the procedure is time-sensitive, with best outcomes if performed before 60 days of age.17 Thus, patients suspected of BA often undergo invasive liver biopsy/cholangiography in order to obtain a clear diagnosis. However, the similarity in presentation between nDJS and BA, and the importance of a successful differential diagnosis, has not attracted much attention and is not a topic discussed in the guidelines for evaluating cholestatic jaundiced infants by the North American and European Societies for Pediatric Gastroenterology, Hepatology, and Nutrition.18 The lack of diagnostic differentiation often leads to nDJS patients being sent for liver biopsy to exclude BA, as shown in our study cohort and in a previous Japanese study cohort.11 To address that issue, we assembled the largest cohort of nDJS patients reported to date at our center to characterize their clinical symptoms, liver chemistries, and pathologic and genetic features. We investigated whether a convenient, noninvasive, biomarker for the early diagnosis of nDJS, differentiating it from BA, could be found. We report here that the liver biochemistry indices of nDJS were generally lower than those in BA. ROC curve analysis of various enzymes revealed that serum ALT and AST levels were robust differentiators of DJS and BA in neonates. The use of an AST cutoff of <87 IU/L was particularly helpful in distinguishing DJS from BA (13/15 nDJS vs. 0/53 BA patients). The finding has potential implications for clinical practice, as liver chemistries are routine and rapid. We believe that such information may be particularly useful in community clinics and nontertiary centers to spare patients invasive procedures such as liver biopsy or cholangiography for diagnosis when nDJS is suspected.5 This study found that an invasive cholangiography can be avoided for the differentiation of BA in patients with cholestasis, acholic stools and AST levels consistently <80 IU/L. We recognize that elevation of ALT and AST can be affected by a variety of factors, including infection, alcohol, and drugs.19,20 AST levels might also be affected by an improper blood collection process or hemolysis. Caregivers must pay attention to those factors when using ALT and AST as biomarkers for the diagnosis of nDJS. Only Chinese patients were enrolled in this study. It would be of interest to determine if the observations made in this report are seen in children from other ethnic groups. It should be noted that increased urinary excretion of coproporphyrin isomer I is a characteristic feature of adult DJS21 and has been proposed by Junge and Norman10 as a potential biomarker to differentiate nDJS from BA. They concluded that urinary coproporphyrin analysis is a fast and reliable diagnostic tool for differentiating nDJS from BA in a comparison of four nDJS patients and twenty-six BA patients. We did not include a urinary coproporphyrin isomer in this study since because we did not have access to a commercial test. The same was true for the 10 nDJS patients reported from the study in Japan.11 Acholic stools are an important diagnostic feature of BA, but their presence in nDJS patients has also been reported by several centers.11,15,22 In a previous study in Japan, eight of ten nDJS patients underwent a liver biopsy because of a suspicion of BA.11 In this study, 13 of the 30 patients, with a described stool color, were reported as having acholic stools, indicating that acholic stool is a relatively common feature of nDJS. Our data suggest that acholic stools are not a reliable differentiator of nDJS and BA. In adult patients, black livers, or melanin-like pigment. are typical histological characteristics of DJS, but not all nDJS patients present with black livers or melanin-like pigment.11,14,23 In this study, no melanin-like pigments were observed in any of the eight patients we biopsied, which indicates that a black liver, or accumulation of melanin-like pigment deposits in hepatocytes, is not a reliable characteristic in nDJS patients. That might be because the infants were too young for accumulation of the pigments, as confirmed by a reported case with prolonged cholestasis in early infancy, and an accumulation of specific pigment granules in the liver cells by 6 years of age.23 The absence of histological accumulation of melanin-like pigment increases the difficulty of diagnosing DJS during the neonatal period. This study reports 33 new DJS variants among the 55 identified in the patients. That substantially expands the current ABCC2 mutation spectrum. Some alleles [c.1177C>T (p.Arg393Trp), c.3825C>G (p.Tyr1275Ter), c.4239_4240dupTC (p.His1414LeufsTer18), c.2755T>A (p.Ser919Thr), c.3928C>T (p.Arg1310Ter), c.298C>T (p.Arg100Ter), c.3196C>T (p.Arg1066Ter), c.2302C>T (p.Arg768Trp), c.1939G>T (p.Glu647Ter)] occurred multiple times, suggesting that they are frequent variants in the population. No significant correlations of patient genotype and the presence of acholic stools were found in the study cohort (Supplementary Material 2). It is reasonable to assume that there exists some other factors affecting the disease phenotype. In addition, we observed some novel clinical features in some patients, including three with transient hypoglycemia and three with transient prolonged prothrombin time (INR), which might be worthy of future investigation. In conclusion, we assembled the largest reported cohort of neonatal Dubin–Johnson syndrome patients that has been evaluated. The most likely negative outcome of their condition is invasive and unnecessary testing to exclude a diagnosis of BA. For that reason, we explored the diagnostic value of liver biochemistry, which is convenient and routine, and found that serum AST level has the potential to act as a sensitive biomarker for the differentiation of suspected nDJS from a possible misdiagnosis of BA in neonates. A formal diagnosis of nDJS would then follow once confirmatory DNA sequencing at the ABCC2 locus was carried out.
Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. |
|
PMC9647113 | Shiwei Wang,Lingling He,Fan Xiao,Meixin Gao,Herui Wei,Junru Yang,Yang Shu,Fuyang Zhang,Xiaohui Ye,Ping Li,Xiaohua Hao,Xingang Zhou,Hongshan Wei | Upregulation of GLT25D1 in Hepatic Stellate Cells Promotes Liver Fibrosis via the TGF-β1/SMAD3 Pathway In Vivo and In vitro | 26-05-2022 | Collagen,Glycosylation,Hepatic stellate cells,Liver fibrosis,TGF-β1 | Background and Aims Collagen β(1-O) galactosyltransferase 25 domain 1 (GLT25D1) is associated with collagen production and glycosylation, and its knockout in mice results in embryonic death. However, its role in liver fibrosis remains elusive, particularly in hepatic stellate cells (HSCs), the primary collagen-producing cells associated with liver fibrogenesis. Herein, we aimed to elucidate the role of GLT25D1 in HSCs. Methods Bile duct ligation (BDL)-induced mouse liver fibrosis models, primary mouse HSCs (mHSCs), and transforming growth factor beta 1 (TGF-β1)-stimulated LX-2 human hepatic stellate cells were used in in vivo and in vitro studies. Stable LX-2 cell lines with either GLT25D1 overexpression or knockdown were established using lentiviral transfection. RNA-seq was performed to investigate the genomic differences. HPLC-MS/MS were used to identify glycosylation sites. Scanning electronic microscopy (SEM) and second-harmonic generation/two-photon excited fluorescence (SHG/TPEF) were used to image collagen fibril morphology. Results GLT25D1 expression was upregulated in nonparenchymal cells in human cirrhotic liver tissues. Meanwhile, its knockdown attenuated collagen deposition in BDL-induced mouse liver fibrosis and inhibited mHSC activation. GLT25D1 was overexpressed in activated versus quiescence LX-2 cells and regulated in vitro LX-2 cell activation, including proliferation, contraction, and migration. GLT25D1 also significantly increased liver fibrogenic gene and protein expression. GLT25D1 upregulation promoted HSC activation and enhanced collagen expression through the TGF-β1/SMAD signaling pathway. Mass spectrometry showed that GLT25D1 regulated the glycosylation of collagen in HSCs, affecting the diameter of collagen fibers. Conclusions Collectively, the upregulation of GLT25D1 in HSCs promoted the progression of liver fibrosis by affecting HSCs activation and collagen stability. | Upregulation of GLT25D1 in Hepatic Stellate Cells Promotes Liver Fibrosis via the TGF-β1/SMAD3 Pathway In Vivo and In vitro
Collagen β(1-O) galactosyltransferase 25 domain 1 (GLT25D1) is associated with collagen production and glycosylation, and its knockout in mice results in embryonic death. However, its role in liver fibrosis remains elusive, particularly in hepatic stellate cells (HSCs), the primary collagen-producing cells associated with liver fibrogenesis. Herein, we aimed to elucidate the role of GLT25D1 in HSCs.
Bile duct ligation (BDL)-induced mouse liver fibrosis models, primary mouse HSCs (mHSCs), and transforming growth factor beta 1 (TGF-β1)-stimulated LX-2 human hepatic stellate cells were used in in vivo and in vitro studies. Stable LX-2 cell lines with either GLT25D1 overexpression or knockdown were established using lentiviral transfection. RNA-seq was performed to investigate the genomic differences. HPLC-MS/MS were used to identify glycosylation sites. Scanning electronic microscopy (SEM) and second-harmonic generation/two-photon excited fluorescence (SHG/TPEF) were used to image collagen fibril morphology.
GLT25D1 expression was upregulated in nonparenchymal cells in human cirrhotic liver tissues. Meanwhile, its knockdown attenuated collagen deposition in BDL-induced mouse liver fibrosis and inhibited mHSC activation. GLT25D1 was overexpressed in activated versus quiescence LX-2 cells and regulated in vitro LX-2 cell activation, including proliferation, contraction, and migration. GLT25D1 also significantly increased liver fibrogenic gene and protein expression. GLT25D1 upregulation promoted HSC activation and enhanced collagen expression through the TGF-β1/SMAD signaling pathway. Mass spectrometry showed that GLT25D1 regulated the glycosylation of collagen in HSCs, affecting the diameter of collagen fibers.
Collectively, the upregulation of GLT25D1 in HSCs promoted the progression of liver fibrosis by affecting HSCs activation and collagen stability.
Liver fibrosis is caused by various chronic liver diseases, can progress to cirrhosis and hepatocellular carcinoma, and it is associated with significant morbidity and mortality.1,2 However, the molecular mechanisms underlying the condition are complex and not fully understood. Apart from etiological treatments, there are no direct antiherpetic fibrosis therapeutics, thus additional research is required. Liver fibrosis characterized by the accumulation of proteins in the extracellular matrix (ECM) following the activation and proliferation of hepatic stellate cells (HSCs).3 Changes in collagen are the most important ECM changes associated with fibrogenesis and fibrinolysis.4 Many post-translational modifications occur during procollagen biosynthesis in the rough endoplasmic reticulum (rER), facilitating proper collagen folding, secretion, and biological functions.5 Of these modifications, proline and lysine hydroxylation are well-recognized and studied.6,7 However, collagen glycosylation has not been extensively investigated. In 1935, Grassmann and Schleich first described collagen glycosylation. Approximately 30 years later, Spiro described the glycan structure: Glc(α1-2) Gal(β1-O) Hyl.8 The corresponding glycosyltransferase enzymes collagen β(1-O) galactosyltransferase 1 (GLT25D1) and collagen β(1-O) galactosyltransferase 2 (GLT25D2) were identified by Schegg et al. in 2009.9 GLT25D1 adds glucosylgalactosyl-hydroxylysine (GG-Hyl) or galactosyl-hydroxylysine (G-Hyl) to specific procollagen hydroxylysine molecules prior to the formation of the triple helix structure in the endoplasmic reticulum (ER).9–11 GLT25D1 regulates cross-linking in bone collagen I.12 Knocking out GLT25D1 in osteosarcoma SaOS-2 cells impaired cell proliferation and viability and induced the upregulation and intracellular accumulation of collagen I.13 In MC3T3-E1 preosteoblast cells, loss of GLT25D1 affects the maturation of collagen cross-linking, fibrillogenesis, and mineralization of collagen fibrils.14 In N-ethyl-N-nitrosourea (ENU)-induced mutant mice carrying a GLT25D1 mutation, collagen IV accumulates inside embryonic fibroblasts and within the ECM.15 GLT25D1 also decreases adiponectin secretion in early obesity and is associated with cerebral small vessel disease through its effect on COL4A1,16,17 indicating that it has a vital role in collagen fibrillogenesis, cross-linking, mineralization, and collagen-cell interaction.12 In light of the findings, we hypothesized that GLT25D1 may play an essential role in activating HSCs, the primary collagen-producing cells associated with liver fibrosis. We used GLT25D1-knockout (KO) mice (C57BL/6J) to investigate this role. However, knocking out GLT25D1 was lethal to the embryos.18 Therefore, heterozygous GLT25D1+/− mice, obtained using Cre-loxP conditional knockout technique, were used. Modeling studies suggest that GLT25D1 may play a role in acute hepatic injury and liver fibrosis.18,19 GLT25D2 deficiency contributes to lipodystrophy and promotes NAFLD.20 We previously showed that GLT25D1 has an essential role in liver disease, although its role and underlying mechanism in liver fibrosis are still poorly understood. In this study, we demonstrated the role of GLT25D1 in primary mouse HSCs and LX-2 cells for the first time. We found that GLT25D1 was upregulated in activated HSCs and played an essential role in HSC activation. GLT25D1 also potentially affected collagen stability. The results suggest that GLT25D1 is a novel and potential antifibrotic target.
Liver samples were collected from eight patients with chronic hepatitis B who underwent hepatic surgery or liver puncture biopsy at Beijing Ditan Hospital between June and November 2020. Three of the patients had mild chronic portal inflammation (S0–S1, control) and five patients had cirrhosis. All biopsy specimens were histologically scored via blinded assessment using the Ishak system.21 Patient baseline information is shown in Supplementary Table 1.
The experimental Animal Welfare Committee of Capital Medical University approved all animal procedures. We used GLT25D1+/− mice, which were generated as previously described.18,19 6–8-week-old male C57BL/6J wild-type (WT) and GLT25D1+/− mice were anesthetized via intraperitoneal injection with 0.8% pentobarbital. After a scalpel midline laparotomy, the liver lobules were turned down and the intestines exteriorized carefully to expose the bile duct. The common bile duct was ligated twice using 6–0 silk sutures and the abdomen was then closed. A sham operation was performed in a similar manner, except that the bile duct was not ligated. The animals were sacrificed on postoperative day 21 and the plasma and liver samples were collected for further experiments.
The total collagen content was assessed by measuring hydroxyproline levels in liver tissue using the kit (Nanjing Jiancheng Bioengineering Institute, Nanjing, China).
The concentration of insoluble collagen (cross-linked collagen) in liver tissue was measured using Biocolor Sircol insoluble collagen assay kits (Biocolor Ltd., Carrickfergus, UK).22 Briefly, 20–30 mg (wet weight) of liver samples were crushed in the fragmentation reagent, incubated at 65°C for 2–3 h, and the supernatant was collected and stained for 30 m using Sircol dye reagent. An acid salt wash was used to remove unbound dye and an alkali reagent was then added to release the collagen-bound dye into the solution using a vortex mixer. Absorbance was measured at 550 nm within 2–3 h of the reaction.
Primary mouse hepatic stellate cells (mHSCs) were isolated from 8–10-week-old male C57BL/6J WT and GLT25D1+/− mice by enzymatic digestion and OptiPrep density gradient centrifugation.23 Briefly, each liver was first perfused in situ through the hepatic portal vein using pronase, DNase I, and type IV collagenase. The liver was then excised, homogenized, and digested in external digestive fluid. The mixture was filtered through a 100 µm mesh and centrifuged at 450×g for 7 m. The supernatant was discarded and the cell pellet resuspended in 5 mL of 15% OptiPrep (Sigma-Aldrich, St Louis, MO, USA) and slowly loaded with 5 mL of 11.5% OptiPrep and then 5 mL Gey’s balanced salt solution (GBSS). After centrifuging at 1,400×g and 4 °C for 17 m, the mHSC layer (white flocculus) between the 11.5% OptiPrep and the GBSS was transferred into a new tube, mixed with GBSS, and centrifuged at 450×g for 7 m. The cell pellet was resuspended in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Primary hepatic stellate cells from two mice of the same group were pooled as one sample. The purity of primary HSCs was determined by adding Oil Red O staining to freshly isolated primary cells that had been cultured for 2 days (>97%).24 HSC viability was determined using the Trypan blue exclusion test (>90%). Isolated mHSCs were cultured in DMEM supplemented with 10% FBS at 37°C in a 5% CO2 atmosphere. Activated mHSCs were obtained 7 days after the initial culture.
Human and mouse liver tissues were fixed in 10% neutral buffered formalin and embedded in paraffin and then divided into 4 µm thick sections. Immunohistochemistry assays were conducted using the liver tissue sections. Tissue slides were deparaffinized, hydrated, incubated with 10% H2O2 for 10 m and then steamed in 0.01 mol/L citrate sodium buffer for 30 m. The reaction was blocked by adding 2% bovine serum albumin (BSA) and incubating for 1 h at room temperature. After blocking, a 1:100 dilution of anti-rabbit GLT25D1 antibody was added and the reaction incubated overnight at 4°C. The following day, the slides were washed three times with phosphate buffered saline (PBS) and then incubated with goat anti-rabbit antibody for 1 h at room temperature. Immunoreactive signals were visualized for 5 m using 3,3′-diaminobenzidine. The cells were grown on 25 mm sterile cell climbing slices for 7 days and then fixed in 4% paraformaldehyde for 15 m at room temperature. After washing three times with PBS, the cells were incubated with 5% BSA for 1 h and then with primary antibody (mouse anti-collagen type I 1:200; Servicebio, Hubei, China) overnight at 4°C. After washing with PBS, the cells were incubated with secondary antibody (CY3; Servicebio, Hubei, China) for 1 h at room temperature. Cell nuclei were stained using diamidino-phenylindole (Biotium, Hayward, CA, USA). Images were acquired using the same microscope parameters (Nikon, Tokyo, Japan) and analyzed using ImageJ software.
SHG/TPEF microscopy is a novel optical tissue imaging system that can be used to characterize fibrillar collagen and analyze pathology-relevant collagen architectural features.25,26 Four micron-thick unstained sections of mouse liver tissues were anonymized and imaged at HistoIndex Pte Ltd (Singapore). Four morphological features were investigated: the number of short strings (length ≤20 µm), the number of long strings (length >20 µm), the number of thin strings (axis ratio ≤0.25), and the number of thick strings (axis ratio >0.25).
Briefly, the lentiviral expression vectors GLT25D1-Plvx-shRNA2-Zsgreen-T2A-puro (Fig. 1E) and GLT25D1-pCDH-CMV-MCS-EF1-GFP-T2A-Puro (Fig. 2A) were successfully constructed at Generay biotech (Beijing, China) and confirmed by sequence analysis. Lentiviruses containing these vectors were prepared and transfected into LX-2 cells using FuGENE 6 transfection reagent (Roche, Basel, Switzerland). Then, single cell-derived clones stably suppressing GLT25D1 (sh-GLT25D1), overexpressing GLT25D1 (OE-GLT25D1), and corresponding negative controls (NC) were selected by subjecting the cells to 2 µg/mL puromycin treatment for one week. Successful section was confirmed using western blot and Quantitative real-time polymerase chain reaction (qPCR).
Two biological replicates from each mouse group (WT and GLT25D1+/−) and three biological replicates from each clonal group (OE-NC and OE-GLT25D1) were pooled and sequenced on the Illumina NovaSeq 6000. Differential gene expression between the two groups was analyzed using the DESeq2 R package (1.20.0). padj ≤ 0.05 and |log2 (fold change)| ≥1 were set as thresholds for significant differential expression. The ClusterProfiler R package (3.8.1) was used to conduct gene ontology (GO) enrichment analysis of the differentially expressed genes.
The human hepatic stellate cell line LX-2 (Institute of Infectious Diseases, Beijing Ditan Hospital) was cultured in DMEM (Gibco, Waltham, MA, USA) with 10% FBS and 1% penicillin/streptomycin (Solarbio, Beijing, CN) at 37°C in a humidified atmosphere containing 5% CO2. The LX-2 cells were starved in serum-free DMEM for 12 h prior to commencing the experiments. Transforming growth factor β1 (TGF-β1, 10 ng/mL), a robust fibrosis mediator, was used to stimulate LX-2 cells to simulate liver fibrosis formation. The control group was treated with the same volume of vehicle. Specific signal pathway inhibitors: SIS3 and SD208 (MedChemExpress, Monmouth Junction, NJ, USA), were added 1 h before adding TGF-β1.
Cell proliferation was measured using the Cell Counting Kit-8 (Dojindo, Kumamoto, Japan) assay following the manufacturer’s protocol. Briefly, LX-2 stable and control cells were seeded in 96-well plates at 2–5×103 cells per well for 12 h and then induced with 10 ng/mL TGF-β1. Thereafter, 10 µL of CCK8 solution was added to each well. Cell growth curves were generated based on optical density absorbance measured at 450 nm using a Varioskan Flash spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).
A wound healing cell migration assay was used to study directional cell migration in vitro.27 Cells were seeded in 12-well culture dishes and cultured for 24 h. The cells were then treated with 10 ng/mL TGF-β1 and scratched with a 10 µL pipette tip. Images were captured under a microscope immediately following wound creation (0 h) and photos were taken every 24 h for 3 days. Images were analyzed in ImageJ and the percentage wound migration area determined.
A standard assay kit (Cell Biolabs, San Diego, CA, USA) was used. LX-2 cells were harvested and resuspended in DMEM. Two parts of the cell suspension was mixed with eight parts of the collagen gel lattice mixture and plated for 1 h at 37°C. After gel polymerization, 1 mL of the medium was added and the mixture incubated for 2 days. Next, 1 mL medium containing TGF-β1 was added and the gels released from the sides of the wells. Images were obtained after 24 h. The changes in collagen gel sizes were analyzed using ImageJ software and normalized to the area of the well.
Total RNA was extracted from cells or tissues using an animal tissue cell total RNA extraction kit (BioSci, Beijing, China), following the manufacturer’s protocol. cDNA was prepared using a PrimeScript RT reagent kit (Takara, Kyoto, Japan), following the manufacturer’s instructions. qPCR was performed using Fast SYBR Green PCR Master Mix (Applied Biosystems, Waltham, MA, USA). Transcript levels were analyzed using the ΔΔCT method and normalized to that of the internal control gene, glyceraldehyde-3-phosphate dehydrogenase. Each reaction was performed in triplicate and each experiment was performed at least three times. The primers used in this study are listed in Supplementary Table 2.
Cells or mouse liver tissues were lysed using RIPA buffer supplemented with protease inhibitor (Roche) and phosphatase inhibitor cocktails 2 and 3 (Sigma). Protein concentrations were measured using a bicinchoninic acid assay kit (Thermo Scientific, Waltham, MA, USA). Proteins were separated on a 10% gel and transferred onto 0.2 µm polyvinylidene difluoride membranes (SDS-PAGE; Millipore, Darmstadt, Germany). After blocking with 5% skimmed milk and washing with TBS-Tween buffer, the membranes were incubated overnight at 4°C with the corresponding antibodies (Supplementary Table 3). The membranes were then incubated with a horseradish peroxidase-conjugated secondary antibody (1:5,000) (ZSGB-bio) for 1 h at room temperature and then with an enhanced chemiluminescence substrate (Millipore, Darmstadt, Germany). Relative protein quantities were determined using ImageJ software. Information on the antibodies is presented in Supplementary Table 3.
Proteins were extracted from lysates of activated sh-GLT25D1, OE-GLT25D1, and control cells. Type I collagen was separated by SDS-PAGE on 10% Bis-Tris precast gels. After staining with Coomassie SimplyBlue, the target protein band was excised from the gel using a scalpel. The bands were cut and diced into 1 mm3 cubes, digested with trypsin, and subjected to mass spectrometry analysis. HPLC-MS/MS analyses were conducted using a RIGOL L-3000 HPLC system (RIGOL, Beijing, China). The MS scan was obtained over a 500−2,000 m/z range with a 2 Hz frequency in positive ion mode. Peptides were identified by searching the acquired MS/MS spectra against Homo sapiens database using the Proteome Discoverer 2.4 software.
Specimens were sampled from the same areas of the liver and fixed using an electron microscope fixation liquid (Solarbio, Beijing, China). The samples were then post-fixed in potassium ferrocyanide-reduced osmium for 1 h at room temperature. Collagen fibrils were observed under a scanning electron microscope. The diameters (µm) of collagen fibers were measured in three different fields of view using Image-Pro Plus software (version 6.0; Media Cybernetics Inc., Rockville, MD, USA).
Data were reported as percentages (%) or means±SD of at least three independent experiments. Statistical analyses were performed using Prism 9 software (GraphPad, La Jolla, CA, USA). Pairwise comparisons between groups were performed using unpaired Student’s t-test. The χ2 test was used to analyze pathological characteristics. For p-values, <0.05, <0.01, and <0.001 were considered statistically significant.
We initially investigated GLT25D1 expression in human liver tissue. α-SMA and GLT25D1 levels were significantly elevated in cirrhotic livers compared with the control livers without fibrosis (Fig. 3A–C). GLT25D1 and α-SMA, a marker of activated HSCs in fibrotic livers, were mainly expressed in nonparenchymal cells (Fig. 3A), suggesting a potential role for GLT25D1 in liver fibrosis and HSCs.
We used heterozygous GLT25D1+/− and WT mice subjected to BDL-induced liver fibrosis to further investigate the in vivo role of GLT25D1 in liver fibrosis. Sirius Red staining and immunohistochemical analysis of α-SMA showed that GLT25D1+/− mice had reduced collagen deposition compared with WT mice (Fig. 4A), in line with decreased hydroxyproline levels (Fig. 4B). The expression of fibrotic genes, including COL1A1, COL3A1, and ACTA2, was significantly downregulated in GLT25D1+/− mice compared with WT mice (Fig. 4C). The GLT25D1 protein levels in liver was provided in the Supplementary Figure 1. The results demonstrate that GLT25D1 knockdown attenuates BDL-induced liver fibrosis in vivo.
IHC results from the analysis of human and mouse liver tissues indicated that GLT25D1 was mainly expressed in nonparenchymal cells. HSCs, which are non- parenchymal cells, are the primary collagen-producing cells and key effectors of liver fibrosis. Therefore, based on the hypothesis that GLT25D1 regulates HSCs in liver fibrosis, we isolated primary HSCs from WT and GLT25D1+/− mouse livers. Quiescent mHSCs showed characteristic lipid droplets (Fig. 5A). mHSCs were activated when the cells were cultured for 9 days, resulting in a spindle-shaped or dendritic morphology (Fig. 5B, left panels). mHSC from WT mice showed more obvious stress fiber-like structures compared with GLT25D1+/− mice after 9 days of culture (Fig. 5B, right panels). To further explore the effects of GLT25D1 on HSCs, RNA-seq analysis of mHSCs from WT and GLT25D1+/− mice was performed after culturing the cells for 12 days. A total of 360 significantly downregulated genes and 22 significantly upregulated genes were identified (Fig. 5C). The expression of fibrosis-related genes was significantly lower in mHSCs from GLT25D1+/− mice compared with those from WT mice, including transforming growth factor beta (TGFβ1), matrix metallopeptidase 9 (MMP9), and MMP12. GO analysis suggested that the genes that were differentially expressed between GLT25D1+/− and control WT mice were associated with cell activation and corresponding downstream profibrogenic responses, including cell chemotaxis and activation (Fig. 5D).28 Immunofluorescence analysis showed that knocking down GLT25D1 reduced intracellular COL I protein levels (Fig. 5E, F). Western blot and RT-PCR results also confirmed that knocking down GLT25D1 in mHSCs significantly decreased α-SMA and intracellular COL I protein levels (Fig. 5G, H). Taken together, these results show that GLT25D1 may be involved in mHSC activation and ECM production.
To further assess the effects of GLT25D1 on HSCs, the human hepatic stellate cell line LX-2, which is commonly used in in vitro hepatic fibrosis studies, was employed. Primary HSCs and LX-2 cells share high gene expression similarities (98.7%).29,30 Western blot analysis revealed that the fibrosis-associated proteins COL I, α-SMA, and GLT25D1 protein were upregulated in activated LX-2 cells (Fig. 1A). qRT-PCR also showed that COL1A1, ACTA2, and GLT25D1 mRNA levels were significantly elevated (Fig. 1B–D). The results demonstrate that GLT25D1 expression was upregulated at both mRNA and protein levels following the TGF-β1-induced activation of LX-2 cells. Thus, GLT25D1 may be involved in LX-2 cell activation and ECM production. Additionally, we constructed a lentivirus vector expressing shGLT25D1 (Fig. 1E) and generated a stable cell line (sh-GLT25D1, Fig. 1F) that successfully knocked down GLT25D1 gene expression (63.2%) than control (Fig. 1G, H). HSC activation is accompanied by specific phenotypes, including proliferation, contractility, and migration.28 The wound healing assay indicated that knocking down GLT25D1 inhibited LX-2’s migratory capacity (p<0.05, Fig. 1I, J). The CCK-8 assay showed that the viability of sh-GLT25D1 cells decreased significantly compared with that of sh-NC cells (Fig. 1K). Collagen gel contraction was used to determine whether GLT25D1 affects HSC contractility. The results showed that knocking down GLT25D1 repressed TGF-β1-induced LX-2 contraction (p<0.05, Fig. 1L), indicating that suppressing GLT25D1 may affect LX-2 cell activation.
Based on the results, we postulated that GLT25D1 is involved in LX-2 cell activation. Rescue experiments were performed to confirm these results by testing whether GLT25D1 overexpression has opposite effects in LX-2 cells, including increased cell proliferation, contractility, and migration. We constructed a GLT25D1 overexpressing lentivirus vector (Fig. 2A) and generated cell lines that stably overexpress GLT25D1 (OE-GLT25D1, Fig. 2B), showing high expression of GLT25D1 protein (260%) and mRNA (173%) compared with the OE-NC control (Fig 2C, D). The results showed that GLT25D1 overexpression increased migratory capacity and promoted cell proliferation (Fig. 2E–G) in OE-GLT25D1 compared with the control. GLT25D1 overexpression also enhanced LX-2 cell contraction (Fig. 2H). Taken together, these results confirm that GLT25D1 may mediate LX-2 cell activation.
To elucidate the effects of GLT25D1 overexpression in LX-2 cells, RNA-seq analysis was conducted in OE-GLT25D1 and OE-NC cells 48 h after their stimulation with TGF-β1. The analysis identified 512 upregulated and 167 downregulated genes (Fig. 6A). The expression of fibrosis-related genes, including COL1A2, COL3A1, ACTA2, TIMP3, PDGFRA, COL11A1, CAV1, and IGFBP3 was significantly upregulated in OE-GLT25D1 cells compared with OE-NC cells. GO enrichment analysis was performed to further explore the biological functions of the DEGs. GLT25D1 greatly affected ECM and collagen formation (Fig. 6B). Western blot and RT-PCR verified the accuracy of the transcriptome data. Western blot results showed that the expression of α-SMA (a marker of LX-2 activation), COL I, COL III, and TIMP-1 were suppressed in sh-GLT25D1 (Fig. 6C-G) but activated in OE-GLT25D1 cells (Fig. 6H-L). Expression of the corresponding liver fibrogenic genes: COL1A2, COL3A1, ACTA2, and TIMP-1, showed similar patterns in sh-GLT25D1 (Fig. 6M–P) and OE-GLT25D1 cells (Fig. 6Q-T) 48 h after they were stimulated with TGF-β1. Taken together, these results indicate that GLT25D1 is required, to some extent, for the activation and expression of fibrogenic genes in LX-2 cells.
The canonical TGF-β/SMAD pathway and noncanonical pathway (MAPK, PI3K/AKT) were evaluated to study the mechanisms of action of GLT25D1. GLT25D1 knockdown decreased TGFβ1-stimulated SMAD3 phosphorylation (Fig. 7A), whereas GLT25D1 overexpression enhanced TGFβ1-stimulated SMAD3 phosphorylation (Fig. 7B). There was no significant difference in noncanonical pathway. Thus, we hypothesized that GLT25D1 regulated activation and ECM production of LX-2 cells via the TGF-β1/SMAD3 signaling pathway. Smad3 phosphorylation inhibitor (SIS3) and TGF-βRI (ALK5) inhibitor (SD208) were used to confirm whether GLT25D1 regulated LX-2 activation by arresting the TGF-β/SMAD3 signaling pathway.31 Western blot analysis showed that SIS3 and SD208 significantly suppressed GLT25D1-induced collagen synthesis and completely abrogated collagen upregulation (Fig. 7C). The results suggest that GLT25D1 promotes the expression of fibrogenic proteins during liver fibrosis through the TGF-β1/SMAD3 signaling pathway.
GLT25D1 adds the monosaccharide Gal(β1-O) to procollagen’s hydroxylysine residue. We investigated whether GLT25D1 influenced specific collagen post-translational modification of COL I in mHSCs and LX-2 cells. In the α1 chain isolated from WT mouse-derived mHSCs, the glycopeptides α1(501–541) were shown to contain two hydroxylated Lys: Hyl-527 and Hyl-541. The Hyl-527 residue was glycosylated by the G-Hyl glycoforms (Fig. 8A). However, the G-Hyl and GG-Hyl glycoforms were not observed in the α1 chain isolated from GLT25D1+/− mHSCs. The α1 chain in LX-2 cells contained three glycosylation sites: at residues α1-586, α1-594, and α1-862, while the α2 chain contained one glycosylation site at residue α2-657. However, only the α1 chain was glycosylated at site α1-586 in sh-GLT25D1 cells. Three additional glycosylation sites were observed in the OE-GLT25D1 α1 chains: at residues α1-448, α1-781, and α1-934 (Supplemental Table 4). The study results confirmed that the glycosylation of collagen from HSCs was regulated by GLT25D1. Post-translational modifications often affect either the molecular weight or the charge of the protein. In this study, the glycosylation changes did not result in an apparent change in the electrophoretic bands (Fig. 8B). To explore whether the glycosylation affected the properties of the fibrotic ECM, collagen fiber diameters and insoluble collagen (cross-linked collagen) were investigated in BDL-induced liver fibrosis in WT and GLT25D1+/− mice. Collagen fibrils in the portal area viewed by SEM (Fig. 8C). After analyzed by Image-Pro Plus software, we found that heterozygous mice had fibers with larger diameters compared with WT mice (Fig. 8D). Insoluble collagen did not differ significantly between heterozygous knockdown and WT mice (Fig. 8E). SHG/TPEF imaging can characterize the architectural features of fibrosis at the individual collagen fiber level.25 We used SHG/TPEF microscopy to capture morphological differences in liver fibrosis between WT and GLT25D1+/− mice. GLT25D1+/− mice had more long strings than WT mice. The results confirm that GLT25D1 regulated glycosylation in type I collagen and influence collagen properties.
We previously showed that GLT25D1 was associated with liver injury, autoimmune liver disease, and nonalcoholic steatohepatitis (NASH).19,20 However, little is known about the role of GLT25D1 in liver fibrosis and HSCs, which are mainly myofibroblast progenitor cells, and the central collagen-producing cells and key effectors in liver fibrosis.26,31 To our knowledge, this is the first study to systematically identify the specific biological functions of GLT25D1 in liver HSCs. We found that GLT25D1 was upregulated in activated HSCs and played an essential role in HSC activation, ECM production, and collagen stabilization. HSCs reside in the space of Disse, between parenchymal cells and endothelial cells (EC) of the hepatic lobule.32 In this study, GLT25D1 and α-SMA (a marker of activated HSCs) were expressed mainly in nonparenchymal liver cells in fibrotic livers. Several studies have reported that GLT25D1 is involved in collagen formation in MC3T3-E1, MEFs, SaOS-2 and collagen-producing cells.13–15 Therefore, we concluded that GLT25D1 expression was also present in activated HSCs. We previously demonstrated that the concentration of GLT25D1 in serum from patients with cirrhosis was significantly higher than in serum from patients without cirrhosis.33 The study results showed that GLT25D1 expression was upregulated in IHCs the cirrhotic liver. Importantly, partial deletion of GLT25D1 significantly reduced BDL-induced collagen deposition and α-SMA expression, indicating that GLT25D1 has an important role in liver fibrosis and HSC activation. HSC activation leads it to transdifferentiate into MFBs, characterized by expression changes and specific phenotypes, including proliferation, contractility, migration, fibrogenesis, matrix degradation, chemotaxis, and inflammatory signaling.34–36 In this study, GLT25D1 deficiency suppressed mRNA and protein expression of specific phenotype- and fibrosis-related genes in both LX-2 and mHSCs. Interestingly, GLT25D1 overexpression in LX-2 cells enhanced gene expression and the phenotypes, illustrating that GLT25D1 is required to activate HSCs. Activated HSCs within the space of Disse migrate toward sites of inflammation, and deposit excessive ECM during the progression of liver fibrosis.37,38 Therefore, enhanced migration is an important aspect of GLT25D1 that contributes to liver fibrosis. Meanwhile, GLT25D1 upregulation affects cell proliferation and contractility. The high number of activated HSCs and contractibility of myofibroblasts in the fibrotic liver promote the constriction of hepatic sinusoids, resulting in blood flow and nutrient exchange.39 Clinically, HSC contraction causes and aggravates portal hypertension that predominantly determines liver stiffness.40,41 In this regard, GLT25D1 may be directly involved in portal hypertension formation, hence indirectly influencing liver stiffness, which correlates with liver fibrosis severity. Therefore, under conditions of chronic inflammation, GLT25D1 upregulation accelerates the progression of liver fibrosis. Baumann et al. reported that the loss of GLT25D1 in SaOS-2 cells induced high expression and accumulation of collagen I in the ER, although this did not result in ER stress.13 Geister et al. found that, in MEFs the loss of GLT25D1 induced the accumulation of type I collagen in the cell but led to its decrease in the culture medium.15 We obtained different results in this study. For both sh-GLT25D1 and mHSCs from GLT25D1+/− mice, COL I levels were significantly suppressed in the cytoplasm. This discrepancy may be due to differences in cell types, the clinical stage of the disease, the levels of GLT25D1 protein, and the knockdown and overexpression methods used. Nevertheless, it is imperative to emphasize that, in the current study, similar results were obtained using mHSC and LX-2. GLT25D1 overexpression induces the production of large quantities of collagen that need proper folding prior to secretion into the extracellular space, challenging the cell’s ER folding capacity. However, it is not clear whether COL I was correctly folded and secreted. Previous studies on GLT25D1 have focused on the structure and secretion of collagen. However, the regulation of signal pathways by GLT25D1 has not been explored. An important finding in this study is that GLT25D1 mediates the activation of LX-2 cells by enhancing the activation of the canonical TGF-β1/SMAD3 pathway. This suggests that GLT25D1, either directly or indirectly, enhances the phosphorylation of Smad3. TGF-β1-induced HSC activation is involved in several signaling pathways, including the canonical TGF-β1/SMAD and non-SMAD-dependent TGF-β signaling. Non-SMAD-dependent TGF-β signaling includes the MAPK, mTOR, PI3K/AKT, and Rho/TPase pathways.42 The canonical TGF-β1/SMAD pathway, where SMAD3 is phosphorylated at the C-terminus, is the main fibrogenic pathway. GLT25D1 is localized in the ER43 while the TGF-β1/SMAD signaling pathway operates in the membrane, cytoplasm and nucleus.42 thus, theoretically, there is no a direct interaction between GLT25D1 and proteins in the TGF-β1/SMAD pathway (TGFβRI, TGFβRII, Smad2/3/4). It is worth noting that collagen, which is regulated by GLT25D1, is also an important molecule mediating HSC activation. Col I binds to the integrin α11β1 to affect cell function.44 However, whether the process involves the TGF-β1/SMAD signaling pathway remains undetermined. Knocking down Gal-3(galectin-3) was correlated with the downregulation of GLT25D1 in CMT cells.45 Gal-3(galectin-3) regulates proliferation and migration of human pulmonary artery smooth muscle cells via the TGF-β1/Smad2/3 signaling pathway.46 Therefore, Gal-3 may participate in the pathway regulated by GLT25D1. Cellular communication network factor 1 (CCN1), which is secreted by fibroblasts, is glycosylated by GLT25D1,47 thus its function is affected by GLT25D1. CCN1 regulates the expression of genes associated with fibrosis in lung fibroblasts and is dependent on the TGF-β1/Smad3 signaling pathway.48 This suggests that CCN1 also participates in the pathway regulated by GLT25D1 in LX-2 cells. The possible mechanism diagram was shown in Figure 9. We confirmed that GLT25D1 regulates glycosylation (GG-Hyl, G-Hyl) in COL I from mHSCs and LX-2 cells by mass spectrometry (Supplementary Table 4). Terajima et al reported that this glycosylation might regulate cross-link maturation and the growth of collagen fibrils in bone.14 Baumann and Hennet also reported that it affects the kinetics of triple helix formation.13 However, the content of insoluble collagen (cross-linked collagen) did not differ significantly between GLT25D1+/−and WT mice in this study. That may be due partly to the short disease course in our fibrosis model. GLT25D1+/− mice had larger fibrils compared with WT mice. This is consistent with the results of previous studies showing that loss of this glycosylation induces larger mean fibril diameters compared with controls.14,49,50 In general, collagen molecules with a higher degree of glycosylation form smaller fibrils that are involved in collagen stability.50,51 This finding may reflect that the alteration in collagen morphology was, to some extent, regulated by GLT25D1, which may be involved in collagen degradation and the reversal of fibrosis. Thus, GLT25D1 affects liver fibrosis by altering the collagen structure. The study has some limitations. GLT25D1 knockdown in GLT25D1+/− mice is not restricted to HSCs, where the biological function of GLT25D1 in hepatocytes during liver fibrosis should not be ignored18,19 and may be accompanied by a compensatory effect of GLT25D2, another galactosyltransferase enzyme isoform.13 Additionally, the expression of GLT25D1 in other hepatic cells remains unclear, including Kupffer cells, macrophages, cholangiocytes, and liver sinusoidal EC. Data on the role of GLT25D1 in the signaling pathway are fairly limited. When investigating collagen properties, a high purity extract of collagen was required, which need to further improve in the further study. Therefore the role of GLT25D1 in hepatic fibrosis requires further investigation. In conclusion, our data showed that GLT25D1 regulated HSC activation and collagen stability. Additionally, GLT25D1 downregulation may alleviate liver fibrosis. Targeting GLT25D1 may be more feasible and safer than directly targeting TGF-β in liver fibrosis and offers a promising therapeutic target for treating liver fibrosis.
Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. Click here for additional data file. |
|
PMC9647116 | Li Chen,Siwei Xia,Shuqi Wang,Yuanyuan Zhou,Feixia Wang,Zhanghao Li,Yang Li,Desong Kong,Zili Zhang,Jiangjuan Shao,Xuefen Xu,Feng Zhang,Shizhong Zheng | Naringenin is a Potential Immunomodulator for Inhibiting Liver Fibrosis by Inhibiting the cGAS-STING Pathway | 28-04-2022 | Liver fibrosis,cGAS,Naringenin,Inflammation,Hepatic stellate cells | Background and Aims Naringenin is an anti-inflammatory flavonoid that has been studied in chronic liver disease. The mechanism specific to its antifibrosis activity needs further investigation This study was to focused on the cyclic guanosine monophosphate-adenosine monophosphate synthase (cGAS) pathway in hepatic stellate cells and clarified the antifibrosis mechanism of naringenin. Methods The relationship between the cGAS-stimulator of interferon genes (STING) pathway and liver fibrosis was analyzed using the Gene Expression Omnibus database. Histopathology, immunohistochemistry, fluorescence staining, Western blotting and polymerase chain reaction were performed to assess gene and protein expression levels associated with the cGAS pathway in clinical liver tissue samples and mouse livers. Molecular docking was performed to evaluate the relationship between naringenin and cGAS, and western blotting was performed to study the expression of inflammatory factors downstream of cGAS in vitro. Results Clinical database analyses showed that the cGAS-STING pathway is involved in the occurrence of chronic liver disease. Naringenin ameliorated liver injury and liver fibrosis, decreased collagen deposition and cGAS expression, and inhibited inflammation in carbon tetrachloride (CCl4)-treated mice. Molecular docking found that cGAS may be a direct target of naringenin. Consistent with the in vivo results, we verified the inhibitory effect of naringenin on activated hepatic stellate cells (HSCs). By using the cGAS-specific agonist double-stranded (ds)DNA, we showed that naringenin attenuated the activation of cGAS and its inflammatory factors affected by dsDNA. We verified that naringenin inhibited the cGAS-STING pathway, thereby reducing the secretion of inflammatory factors by HSCs to ameliorate liver fibrosis. Conclusions Interrupting the cGAS-STING pathway helped reverse the fibrosis process. Naringenin has potential as an antihepatic fibrosis drug. | Naringenin is a Potential Immunomodulator for Inhibiting Liver Fibrosis by Inhibiting the cGAS-STING Pathway
Naringenin is an anti-inflammatory flavonoid that has been studied in chronic liver disease. The mechanism specific to its antifibrosis activity needs further investigation This study was to focused on the cyclic guanosine monophosphate-adenosine monophosphate synthase (cGAS) pathway in hepatic stellate cells and clarified the antifibrosis mechanism of naringenin.
The relationship between the cGAS-stimulator of interferon genes (STING) pathway and liver fibrosis was analyzed using the Gene Expression Omnibus database. Histopathology, immunohistochemistry, fluorescence staining, Western blotting and polymerase chain reaction were performed to assess gene and protein expression levels associated with the cGAS pathway in clinical liver tissue samples and mouse livers. Molecular docking was performed to evaluate the relationship between naringenin and cGAS, and western blotting was performed to study the expression of inflammatory factors downstream of cGAS in vitro.
Clinical database analyses showed that the cGAS-STING pathway is involved in the occurrence of chronic liver disease. Naringenin ameliorated liver injury and liver fibrosis, decreased collagen deposition and cGAS expression, and inhibited inflammation in carbon tetrachloride (CCl4)-treated mice. Molecular docking found that cGAS may be a direct target of naringenin. Consistent with the in vivo results, we verified the inhibitory effect of naringenin on activated hepatic stellate cells (HSCs). By using the cGAS-specific agonist double-stranded (ds)DNA, we showed that naringenin attenuated the activation of cGAS and its inflammatory factors affected by dsDNA. We verified that naringenin inhibited the cGAS-STING pathway, thereby reducing the secretion of inflammatory factors by HSCs to ameliorate liver fibrosis.
Interrupting the cGAS-STING pathway helped reverse the fibrosis process. Naringenin has potential as an antihepatic fibrosis drug.
Liver fibrosis is a chronic disease generated by liver injuries caused by several factors, such as excessive alcohol consumption, virus infection (including hepatitis B and hepatitis C), nonalcoholic steatohepatitis (NASH), nonalcoholic fatty liver disease, and autoimmune hepatitis.1–3 Multiple types of liver damage and disease have led to the high prevalence and mortality of liver fibrosis in China and worldwide.4 Without appropriate treatment, liver fibrosis gradually results in hepatocellular carcinoma.2 As the driving center of liver fibrosis, hepatic stellate cells (HSCs) are implicated in hepatic inflammation5 and considered a main factor that promotes liver fibrosis.6 HSCs receive signals from leukocytes within the hepatic environment, amplify the signals, and produce molecules that then target and modulate leukocytes. Specifically, HSCs promote leukocyte chemotaxis and adherence, and may also regulate the activation of leukocytes within the hepatic environment by secreting immunoregulatory cytokines.7 Inhibition of HSC-associated inflammatory signals effectively ameliorates liver fibrosis.8,9 However, the specific regulatory mechanism of inflammation in HSCs needs further study. As an inflammation-mediated pathway, the cyclic guanosine monophosphate-adenosine monophosphate synthase (cGAS) stimulator of interferon genes (STING) signaling pathway has been widely studied in recent years for its role in chronic liver disease, including liver fibrosis, which is an important process associated with chronic liver disease. cGAS is an inflammatory pathway activation sensor. When the cGAS-STING signaling pathway is activated, the expression of type I interferon and other inflammatory cytokines are induced, such as interferon regulatory factor 3 (IRF3), and triggers the innate immune response.10 STING is widely expressed in various cell types, and recent studies have found that it can regulate different pathways of programmed cell death. Therefore, a deeper understanding of the cGAS-STING signaling pathway may provide a new method of treating chronic inflammatory diseases.11 Recently, many studies have shown that the cGAS-STING signaling pathway is associated with the occurrence of liver fibrosis.12 Thus, targeting the cGAS-STING pathway may be a potential therapeutic strategy for liver fibrosis. Naringenin is a flavonoid compound, that has been studied in various liver damage models, induced by carbon tetrachloride (CCl4), alcohol, N-methyl-N-nitro-nitroguanidine, lipopolysaccharide (LPS), and heavy metals in vivo and in vitro. In those studies, naringenin had a good hepatoprotective activity because of its antioxidant activity and its ability to inhibit inflammatory and fibrotic signaling pathways.13 Although naringenin has been reported to reduce inflammation and liver fibrosis, the mechanism underlying this action still needs to be fully revealed. Whether the cGAS-STING pathway is involved in inhibiting inflammation and the antifibrotic effect of naringenin has not been reported. Therefore, the aim of the present study was to explore the mechanism underlying the effects of naringenin in liver fibrosis treatment. We showed that naringenin affected the secretion of inflammatory factors and the phenotypic changes of HSCs by interfering with the cGAS-STING signaling pathway to alleviate liver fibrosis.
After downloading the gene expression matrix of GSE99807 and GSE33650 (https://www.ncbi.nlm.nih.gov/geo/), the information and RNA expression data of four patients with liver cancer tissues and pericarcinoma tissues were extracted from GSE99807. The gene expression data of patients with normal liver tissues and HCV infected liver fibrosis tissues were extracted from GSE33650. The selection condition for differentially expressed genes between liver cancer tissues and adjacent normal tissues in GSE99807 was p<0.05. The logFC values of genes of the samples (GSE99807) were used to draw a volcano map. The differentially expressed genes (GSE99807) were used for KEGG signal enrichment analysis and gene ontology biological processes (GOBP) enrichment analysis (https://david.ncifcrf.gov was used to obtain analysis results). TMEM173 and IRF3 gene expression analysis was performed based in the RNA expression data of GSE33650. The data was visualized by using the cloud platform http://sangerbox.com/Tool.
The liver cancer tissues and adjacent normal tissues of human liver cancer patients were collected in Jiangsu Provincial Hospital of Traditional Chinese Medicine from July 2018 to December 2021. According to the inclusion and exclusion criteria, the study collected tumor tissues and adjacent normal tissues from three patients with liver cancer. The inclusion was surgical treatment with pathological diagnosis of liver cancer in tissue specimens. Before this visit, the patient had not received any radiotherapy, chemotherapy, or biological treatment. Patients with autoimmune diseases complicated by primary tumors in other organs other than liver cancer, with tumor metastases in organs other than the liver before and within 6 months after the operation, poor physical condition, and expected postoperative survival of <6 months were excluded.
Naringenin (must-21032406) was purchased from CDMUST (Chengdu, China). Other reagents were obtained from Sigma-Aldrich (St. Louis, MO, USA). Anti-alpha smooth muscle actin (α-SMA) (A7248), anti-cGAS (A8335), anti-IRF3 (A2172), anti-P-TBK1 (AP0847), anti-TBK1 (A3458), anti-IL8(A12452), and anti-STING (A20175) were purchased from ABclonal (Woburn, MA, USA). Anticollagen I (14695-1-AP), anti-β-actin (66009-1-Ig), anti-rabbit IgG (SA00001-2), and anti-mouse IgG (SA00001-1) were purchased from Proteintech Group (Rosemont, IL, USA). Anti-IL1β (ab234437), anti-IL6 (ab229381), and anti-TNFα (ab215188) were purchased from Abcam (Cambridge, UK).
Male C57BL/6J mice weighing 18–20 g were obtained from the Nanjing Qinglongshan Animal Co. Ltd. (Beijing, China). All animals were cared for humanely following National Institutes of Health (Bethesda, MD, USA) guidelines. When conducting experiments on animals, we followed the 3R principle and respected the highest ethical and animal welfare standards. Before the procedures, all animals were kept in an air-conditioned room at 24°C, with a 12-hour dark/light cycle for 1 week. All animals were free to get food and water during the study. Eighteen mice were randomly divided into three groups of six each. Mice in group 1 were intraperitoneally injected with olive oil as the negative control, mice in group 2 were intraperitoneally injected with CCl4 (1:9 v/v with olive oil) at a dose of 5 ml/kg for 8 weeks (3 times/week) to induce liver fibrosis, and mice in group 3 were given intragastric naringenin 100 mg/kg14 for 8 weeks (3 times/week) with CCl4 modeling.
Mouse blood was collected from the eye socket vein, for assay of serum liver biochemical indexes, alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), Laminin (LN), hyaluronic acid (HA), procollagen type-III (PC-III) and collagen type-IV (IV-C) were analyzed using a Hitachi 7020 Chemistry Analyzer (Tokyo, Japan). The inflammatory factors were detected by ELISA kits, interleukin (IL)-1β (YFXEM00028), IL-6 (YFXEM00045), IL18 (YFXEM00573), and tumor necrosis factor alpha (TNFα, YFXEM00031) purchased from Yi Fei Xue (Nanjing, China).
The liver tissue was fixed with 10% neutral buffer formalin and embedded in paraffin. Liver sections with a thickness of 5 µm were prepared and stained with hematoxylin and eosin, and Masson pine trichromatic staining was performed by standard methods. For Sirius red collagen staining, sections were dewaxed, and stained with Sirius red for 1 h at room temperature. After washing, the slides were dehydrated in 100% ethanol and xylene and then mounted in Permount. The photos were taken in a random area blindly.
Immunofluorescence staining of LX2 cells or slides of liver mouse and human tissues was performed according as previously reported.15 Diamidino-phenyl-indole (DAPI) was used to stain the nucleus. In some immunofluorescence staining of liver tissues, cGAS-STING pathway-related molecule (cGAS, 1:100 diluted in 5% bovine serum albumin [BSA]) and activated HSC marker α-SMA (1:100 diluted in 5% BSA) were doubly stained. Pictures were taken in five random areas with a fluorescence microscope (Zeiss, Oberkochen, Germany).
The binding of naringenin to cGAS protein (PDB 6047) was modeled with GLIDE software (Schrödinger, LLC, New York, NY, USA). The guide protein preparation method used with the Maestro workstation was to remove all water molecules from the original structure during protein preparation, adding electric charges and hydrogen atoms, and modify the geometry of all heterogroups separately. After that, the Prime tool was used to predict and fill in missing ring atoms and hydrogen bonds. Finally, IMPREF optimized the placement of hydrogen bonds and keeps all the atoms in place. The default constraints of OPLS_2005 force field and 0.3a of Root Mean Squared Error (RMSD) were used for energy minimization. The receptor mesh generation panel was used to generate grids under the silicon chip target screening, defining the receptor structure to exclude any other primitive compounds that may be present, and the method of settling the location and size of the active site by the receptor mesh was previously reported by Dai et al.16
Human LX2 and L02 cells were obtained from the Cell Bank of Chinese Academy of Sciences (Shanghai, China) and cultured in DMEM (Invitrogen, Grand Island, NY, USA) supplemented with 1% fetal bovine serum and grown in a 5% CO2 humidified atmosphere at 37°C. Dimethyl sulfoxide (DMSO) was selected as the carrier for dissolving naringenin. LX2 and L02 cells were treated with 0, 2, 5, 10, 20, or 40 µM naringenin for 24 h. Cell viability indicative of cellular metabolic activity was measured using MTT assays as described by Wang et al.17 The spectrophotometric absorbance at 490 nm was determined using a SPECTRAmax microplate spectrophotometer (Molecular Devices, Sunnyvale, CA, USA).
Protein extracts were prepared from the liver and LX2 cells using RIPA buffer. Protein detection, WBs, and quantification were performed by standard methods. All primary antibodies were diluted 1:2,000 with 5% BSA, and secondary antibodies were diluted 1:10,000 with 5% skimmed milk powder. β-actin was used as a constant control for total protein equivalent loading.
Trizol reagent (Sigma-Aldrich) was used to isolate total RNA from LX-2 cells. RNA (2 µg) was reverse transcribed by a reverse transcription-polymerase chain reaction kit to obtain cDNA. NanoDrop assays was used to quantify RNA. Primers were purchased from Tsingke Biotech (Nanjing, China). The mRNA of interest was normalized to GAPDH. GAPDH was used as endogenous control. The primers for the target genes were listed in Table 1. The relative expression changes were determined by the 2−ΔΔ CT method.
Values were reported as means and standard deviation. Between-group differences were compared by unpaired student t-tests. Multiple-group differences were compared by one-way analysis of variance with Bonferroni correction (Graph Pad Prism 9.0, San Diego, CA, USA). P-value <0.05 were considered statistically significant.
RNA microarray analysis data from liver cancer and adjacent normal tissues from eight patients showed that the gene expression levels of TMEM173, COL1A1, and IRF3 were higher in cancer tissues than pericarcinoma tissues (Fig. 1A, B). The abnormally expressed genes in liver fibrosis tissues from liver cancer patients were enriched in multiple signaling pathways, including the primary immunodeficiency, and NF-kappa B signaling pathways, which have been reported to promote the development of chronic liver disease. Obviously, the signaling pathways reported by some studies were closely related to the functional regulation of STING in chronic liver disease and liver cancer,18 which may contribute to promoting the development of liver fibrosis (Fig. 1C). The top 12 enriched biological processes derived from the differentially expressed genes between liver cancer tissues and adjacent normal tissues in patients showed that the inflammatory response was involved in the malignant progression of liver fibrosis (Fig. 1D). The analysis based on GSE33650 showed that TMEM173 (STING) and IRF3 were highly expressed in liver fibrosis tissues from liver cancer patients (Fig. 1E, F). The liver fibrosis samples also had high expression of genes related to the cGAS-STING pathway (Supplementary Fig. 1A). The results showed that STING-related inflammatory pathways had an important role in the development of liver fibrosis, and that the inflammatory response signal was evoked along with STING.
Among the samples collected from liver cancer patients, HE, Masson, and Sirius red staining confirmed the occurrence of liver fibrosis (Fig. 2A). The Masson and Sirius red staining results showed that collagen deposition and fibrotic lesions were more extensive in the area of the tumor sample, which represented the liver fibrosis area in the tumor tissue. Compared with adjacent normal tissues, immunofluorescence staining showed that the expression levels of α-SMA and cGAS (Fig. 2B) were significantly increased in liver fibrosis tissues from liver cancer patients. The results showed that the increase in liver fibrosis was accompanied by an increase in cGAS expression. The WB results showed that the protein levels of cGAS and STING were higher in the tumor tissues of patients than the adjacent normal tissues (Fig. 2C). The above results indicate that the occurrence and development of liver fibrosis were associated with the cGAS-STING pathway.
To evaluate whether naringenin alleviated liver fibrosis, we used hematoxylin and eosin, Masson, and Sirius red staining to detect the level of liver damage and fibrosis in the liver tissue of naringenin-treated mice with CCl4-induced liver fibrosis. Serum biochemical indicators (ALT, AST, ALP, ALT, LN, HA, PC-III, and IV-C) in mice were also assayed to determine changes in liver function. The immunohistochemistry results showed that naringenin significantly reduced liver tissue damage and liver fibrosis caused by CCl4 (Fig. 3A). Serological testing showed that naringenin reduced the levels of ALT, AST, ALP, and ALT (Fig. 3B) as well as the levels of LN, HA, PC-III, and IV-C (Fig. 3C) in CCl4-induced liver fibrosis. In addition, the ELISA results showed that naringenin significantly reduced the expression of inflammatory factors, such as IL1β, IL6, IL18 and TNFα in CCl4-induced liver fibrosis (Fig. 3D). The results indicated that naringenin alleviated the symptoms of liver damage and liver fibrosis caused by CCl4 and reduced the inflammatory response induced by chemical damage.
HSC activation is closely associated with inflammation. cGAS is an important link in the process of inflammatory activation. Therefore, we performed fluorescence staining of the HSC activation indicators α-SMA and cGAS in mouse liver tissue sections. The results showed that α-SMA and cGAS were coexpressed in liver tissue and naringenin reduced the expression of α-SMA and cGAS in CCl4-induced liver fibrosis (Fig. 4A). Subsequently, we docked naringenin and the cGAS complex using the molecular modeling packages in Maestro workstation to explore the potential binding mode of naringenin and cGAS protein. cGAS crystal structure was obtained from the RCSB Protein Data Bank (PDB code: 6047). The docking simulation results confirmed that naringenin bound to the hydrophobic pocket and partially overlapped the binding sites of cGAS, which disrupted the dimerization of cGAS (Fig. 4B). The result showed that naringenin fit into the hydrophobic pocket of cGAS. The docking score of the two compounds was −6.179.
We attempted to verify the antihepatic fibrosis effectiveness and mechanism of naringenin in liver fibrosis in vitro. MTT assays showed that naringenin significantly inhibited fibrosis in LX2 cells at 20 µM but had no effect in L02 cells (Fig. 5A). Therefore, in follow-up experiments, naringenin concentrations of 0, 10, 20, and 40 µM were used to verify the dose-dependent influence. The real-time PCR results showed that naringenin inhibited the expression of the LX2 activation markers α-SMA and α1-procollagen in a dose-dependent manner (Fig. 5B). The WB results also showed that naringenin dose-dependently reduced the α-SMA and collagen1 protein in LX2 cells (Fig. 5C). Immunofluorescence staining further confirmed the above results (Fig. 5D). The results indicated that naringenin significantly inhibited HSC activation in vitro.
To explore the mechanisms underlying the ability of naringenin to reduce liver fibrosis, we assayed the mRNA expression of cGAS and related inflammatory factors in LX2 cells by real-time PCR. The results showed that naringenin significantly reduced the expression of cGAS and STING mRNA in LX2 cells (Fig. 6A), and also inhibited the mRNA levels of inflammatory factors, such as IL1β, IL6, NF-kappa B, and IL8 in LX2 cells (Fig. 6B). WB assays confirmed that naringenin dose-dependently reduced the expression cGAS, STING, and related inflammatory factors in LX2 cells (Fig. 6C), and also dose-dependently reduced the expression IRF3 protein (Supplemental Fig. 1B). To clarify the mechanism underlying the effect of naringenin on the cGAS-STING pathway, we added exogenous dsDNA (1 µg/ml) to LX2 cells to specifically stimulate the cGAS signal, and treated LX2 cells with naringenin (40 µM) for 24 h. WB assays showed that dsDNA promoted the secretion of IL1β, IL6, TNF-α, and IL8 by activating cGAS and STING and that naringenin suppressed those effects (Fig. 6D). Naringenin also offset the activation of IRF3 induced by dsDNA (Supplementary Fig. 1C). The results indicate that naringenin inhibited the cGAS-STING pathway in HSCs, thereby inhibiting the secretion of inflammatory factors and promoting the alleviation of liver fibrosis.
Persuasive evidence has indicated that anti-inflammatory drugs block the progressive development of liver fibrosis.19 In this study, an analysis of the GEO database showed that the inflammation-related cGAS-STING pathway may be closely associated with the development of chronic liver disease and fibrosis. In clinical liver tissue samples, we found evidence that cGAS-STING is positively related to the expression of liver fibrosis indicators. In the CCl4-induced mouse liver fibrosis model, we confirmed that naringenin primarily reduced symptoms of fibrosis by inhibiting inflammation. In particular, molecular docking experiments showed that the molecular structure of naringenin allowed binding with cGAS. The combined application of the cGAS-specific agonist dsDNA and naringenin confirmed the specific binding relationship between naringenin and cGAS at the cellular level. Binding inhibited the activation of the cGAS-STING pathway in activated HSCs, thereby reducing the secretion of inflammatory factors and ultimately improving liver fibrosis indicators. Liver fibrosis is considered an inflammation-related disease,20 and fibrotic livers present with a large accumulation of inflammatory factors. Although the emerging concept of immune metabolism mainly focuses on immune cells, HSCs represent a significant contribution to the convergence pathways of metastatic inflammation and tissue injury, especially in alcoholic and nonalcoholic steatohepatitis.8 HSCs produce and respond to inflammatory mediators, including cytokines and chemokines that amplify the liver’s response to injury. The interaction between HSCs and inflammatory cells can drive inflammation and injury through paracrine signals. For example, HSCs interact with macrophages through the MER-TK receptor to cause fibrosis in experimental NASH.21 cGAS is a cytoplasmic pattern recognition receptor that acts as a DNA sensor to activate the NF-kappa B signaling pathway to produce inflammatory factors,22 STING can bind to activate NF-kappa B, which is the main regulator of inflammation and considered to play a central role in liver injury.23 Many studies have confirmed that NF-kappa B is closely related to the progression of liver fibrosis. For example, Liu et al.24 reported that Liuweiwuling tablets improved BDL-induced liver fibrosis, mainly by inhibiting the NF-kappa B signaling pathway. Additional evidence from Liu et al.25 showed that direct inhibition of NF-kappa B signaling pathway activation induced by drugs in activated HSCs inhibited secretion of downstream inflammatory factors, which had an antifibrosis effect in the liver. Many studies have focused on the ability of NF-kappa B to regulate the production of inflammatory factors in HSCs and have shown that inhibiting the NF-kappa B signaling pathway using drugs improves liver fibrosis, but studies on antiliver fibrosis drugs related to cGAS are relatively rare. As an inflammation-mediated molecule that has received more attention in recent years, cGAS may have a key role in the secretion of inflammatory factors in HSCs. Persistent liver damage can cause liver cells to rupture, and their intracellular contents act as damage-related molecular patterns (DAMPs), thus causing additional leukocyte infiltration and amplifying the original damage. Necrosis-derived DNA can be recognized as a DAMP that activates liver nonparenchymal cells, such as HSCs.26 As a DNA sensor, cGAS must play an important role in the progression of liver damage. Naringenin is a flavonoid with good anti-inflammatory activity. In this study, we verified for the first time that naringenin and cGAS can directly combine. Naringenin can be used as a specific antagonist of cGAS to antagonize the effect of dsDNA, which attenuates the secretion of inflammatory factors in HSCs and inhibits the activation of HSCs. The results of this study provide important insights on the anti-inflammatory effects of naringenin and its potential pharmacological mechanism for the treatment of chronic liver disease. In summary, the study shows that naringenin inhibited HSC activation and inflammation by disrupting the cGAS-STING signaling pathway in HSCs, thus preventing the further development of CCl4-induced liver fibrosis in mice. Interruption of the cGAS-STING pathway was helpful in reversing the fibrosis. The results indicate that naringenin may help to prevent or reverse the progression of liver injury and may be used to develop a new therapeutic approach for chronic liver diseases.
Click here for additional data file. |
|
PMC9647118 | Gan Du,Ruizhe Yang,Jianguo Qiu,Jie Xia | Multifaceted Influence of Histone Deacetylases on DNA Damage Repair: Implications for Hepatocellular Carcinoma | 13-09-2022 | Histone deacetylases,DNA repair,Hepatocellular carcinoma | Hepatocellular carcinoma (HCC) is one of the most commonly diagnosed cancers and a leading cause of cancer-related mortality worldwide, but its pathogenesis remains largely unknown. Nevertheless, genomic instability has been recognized as one of the facilitating characteristics of cancer hallmarks that expedites the acquisition of genetic diversity. Genomic instability is associated with a greater tendency to accumulate DNA damage and tumor-specific DNA repair defects, which gives rise to gene mutations and chromosomal damage and causes oncogenic transformation and tumor progression. Histone deacetylases (HDACs) have been shown to impair a variety of cellular processes of genome stability, including the regulation of DNA damage and repair, reactive oxygen species generation and elimination, and progression to mitosis. In this review, we provide an overview of the role of HDAC in the different aspects of DNA repair and genome instability in HCC as well as the current progress on the development of HDAC-specific inhibitors as new cancer therapies. | Multifaceted Influence of Histone Deacetylases on DNA Damage Repair: Implications for Hepatocellular Carcinoma
Hepatocellular carcinoma (HCC) is one of the most commonly diagnosed cancers and a leading cause of cancer-related mortality worldwide, but its pathogenesis remains largely unknown. Nevertheless, genomic instability has been recognized as one of the facilitating characteristics of cancer hallmarks that expedites the acquisition of genetic diversity. Genomic instability is associated with a greater tendency to accumulate DNA damage and tumor-specific DNA repair defects, which gives rise to gene mutations and chromosomal damage and causes oncogenic transformation and tumor progression. Histone deacetylases (HDACs) have been shown to impair a variety of cellular processes of genome stability, including the regulation of DNA damage and repair, reactive oxygen species generation and elimination, and progression to mitosis. In this review, we provide an overview of the role of HDAC in the different aspects of DNA repair and genome instability in HCC as well as the current progress on the development of HDAC-specific inhibitors as new cancer therapies.
Hepatocellular carcinoma (HCC) is the sixth most commonly diagnosed cancer and the third leading cause of cancer-related mortality worldwide. According to the World Health Organization’s estimation, 905,677 new liver cancer cases and 830,180 affected individuals died in 2020. The 5-year survival rate of HCC is 18%, indicating poor prognosis and limited available treatments. Hepatocarcinogenesis and the development of HCC are complex processes with multiple risk factors, including chronic infection with hepatitis B or C viruses (HBV or HCV, respectively), alcoholism, and exposure to dietary aflatoxin. HCC development involves constant inflammation, causing hepatocyte necrosis and regeneration, which is accompanied by fibrotic generation. As a result of genovariation in passengers, driver genes, and epigenetic modifications, HCC exhibits great molecular heterogeneity.1 Genomic instability, which expedites the acquisition of genetic diversity, acts as a facilitating characteristic of cancer hallmarks. Genomic instability is associated with a greater tendency to accumulate DNA damage, which gives rise to gene mutations and chromosomal damage and causes oncogenic transformation and tumor progression.2 More than 10,000 genes have been detected as significantly mutated genes in HCC, and 26 genes were altered most frequently, such as TP53, CTNNB1, and AXIN1.3 The high frequency of mutability caused by DNA damage leads to the selective advantage of subclones of cells in tumor tissue. DNA repair pathways, accounting for cell viability by annealing double-strand break (DSB) sites, are deemed a basic origin of resistance to chemotherapy and radiation therapy. In minute DNA repair pathways, DNA repair inhibitor administration needs to be concentrated on select patients with particular DNA mutations. For example, olaparib possesses precise treatment potential for DNA damage response (DDR)-mutated HCC.4 Taken together, these results emphasize multiple functions of HCC attained from gene mutation and genomic instability. Histone deacetylases (HDACs) have been shown to impair a variety of cellular processes of genome stability, including the regulation of DNA damage and repair, reactive oxygen species (ROS) generation and elimination, and progression to mitosis. Targeting genome integrity in rapidly cycling cells has always been a preferred strategy in cancer therapy;5 in this review, we focus on the different aspects of genome instability induced by pharmacological inhibition of HDACs. Here, we illustrate the main processes of DNA repair and epigenetic modification presented by deacetylation in HCC and discuss the possible relationship between them, with the intention of proposing a novel therapeutic strategy by integrating DNA repair and HDAC inhibitors for HCC administration.
DNA impairment, including single-strand breaks, DSBs, bulky adducts, base alkylation, base mismatches, insertions, and deletions, is caused by various environmental agents, such as cigarette smoke, ultraviolet radiation, industrial chemicals,6 chemotherapy drugs, and intrinsic agents, such as oxygen radicals and metabolites.7 DSBs are recognized as one of the ultimate roots for DNA instability and mutation and are associated with several specific repair mechanisms (Fig. 1). Homologous recombination (HR) and classical nonhomologous end joining (NHEJ) act as the major errorless repairs of DSBs, while alternative end joints (alt-EJs) and single-strand annealing (SSA) operate as backups of NHEJ and HR.
HR, mainly occurring in the S and G2 phases, is a highly conservative and faultless mechanism. (Fig. 2) Its repair involves homologous DNA from the sister chromatid, which, when used as a model, avoids possible mistakes. First, the impaired DNA forms 3′-overhang single-stranded DNA (ssDNA), recruiting human C-terminal-binding protein (CtIP) to bind at the DSB sites as an initiation to enable the MRN complex (constituted by MRE11, RAD50 and NBS1) to attain its nuclease activity and to regulate nucleases EXO1 and BLM/DNA2.8 CtIP, as a sensor for DNA damage, controls MRN-directed resection.9 Phosphorylated RPA loading at ssDNA as a bridge is replaced by recombinase RAD51, which orchestrates breast cancer susceptibility protein (BRCA1)-BRCA1-associated RING domain 1 (BARD1), PALB2, and BRCA2 to make up a helical nucleoprotein filament, facilitating sister chromatid involvement.10 The filament, in order to repair the lesion, may either undergo the synthesis-dependent strand annealing (SDSA) pathway, with engagement of the Holliday junction, or the double-strand break repair (DSBR) pathway, followed by the recruitment of multiple enzymes, such as GEN1, BLM/Top3a/RIM1 and Mus81-Eme1.11
A rapid but not sufficiently accurate mechanism compared with HR, NHEJ mainly occurs in G1 phase, which connects broken DSBs with randomly synthesized nucleobases. (Fig. 3) Ku (Ku7080 heterodimer) first combines with DSBs as a loading protein to recruit DNA-dependent protein kinase catalytic subunit (DNA-PKcs).11 DNA-PKcs and Ku together constitute the Ku/DNA-PKcs complex as DNA-dependent protein kinase (DNA-PK).12 DNA-PKcs undergoes autophosphorylation and then recruits and phosphorylates Artemis.13 Phosphorylated Artemis gains its DNA-PK-dependent 5′ and 3′ endonuclease activity and 5′ to 3′ single-stranded DNA exonuclease activity, enabling it to cut the dissociative DNA end.14 After that, DNA polymerases (pol), including pol λ, pol µ and terminal deoxynucleotidyl transferase (TdT), are involved in ligation.15 Otherwise, DNA-PKcs also regulate the essential DNA ligation module Ligase4/X-ray repair cross-complementing 4 (XRCC4)/XLF to stabilize the DNA end structure and fine-tune DNA end ligation.11
Alt-EJ operates as the backup mechanism of NHEJ. Although alt-EJ can fix DSBs, it will very likely result in large alterations and even the formation of chromosomal translocations.16 Poly(ADP-ribose) polymerase 1 (PARP1) is involved in sensing DNA damage and binding the end of the DNA.17 The MRN complex, which is phosphorylated by CtIP and initiates alt-EJ, can be inhibited by Ku competitive combination with DSBs.18 Alt-EJ can start only if the content of Ku hovers at a relatively low level. The MRN generates 15- to 100-nucleotide 3′ overhangs through its endonuclease function, exhibiting the microhomology of DSBs, where the DNA pol θ extends the DNA ends, utilizing the opposite DNA sequence as a replication template.19 The stable annealing partner is ultimately sealed by DNA ligase I or DNA ligase III (Fig. 4).20
SSA, as a backup to HR, is prone to induce mutations along with severe deletions and translocations.21 Although HR is the dominant repair mechanism under normal conditions, SSA exerts its function when HR-dependent proteins, RAD51, and its mediator proteins, such as BRCA2 and RAD54, are disrupted.22 MRN and CtIP are involved in creating the 3 DNA tails, and then EXO1, BLM, and DNA2 extend the tails.23 Multiple copies of RPA combine with the prolonged DNA end for stability and protection, lessening the formation of secondary structures.24 After that, RAD52 substitutes for RPA for homology search, strand invasion and annealing.25 Furthermore, the redundant unannealed flaps are removed by the ERCC1/XPF nuclease, and possible gaps are filled by DNA ligase1 (Fig. 5).26 CtIP and MRE11 act as the collective basic molecules of HR, alt-EJ and SSA to start these pathways, whereas NHEJ is initiated by its unique starter, Ku. DNA end resection is of vital importance to pathway choice. The unfavorable environment for resection strengthens the stability of Ku70-Ku80, leading to an inclination of NHEJ. Dislodgement of Ku70–Ku80, as well as the appearance of long-range resections, turns the repair into HR. The error-prone pathways alt-EJ and SSA can hijack the normal HR pathway and generate chromosomal rearrangements.27 p53-binding protein 1 (53BP1) binds to DNA ends and form irradiation-induced foci, limiting the length of resection and prompting NHEJ.28 The function of the Shieldin complex is similar to that of 53BP1, blocking DNA end resection and inducing NHEJ.29 Additionally, phosphorylase ataxia-telangiectasia mutated (ATM) and ataxia telangiectasia and Rad3-related protein (ATR) can activate various HR factors, such as MRN, CtIP and EXO1, and enhance DNA end resection-related pathways.30 Moreover, BRCA2 and RAD51 can overcome the resistance of 53BP1 and the Shieldin complex toward DNA end resection and recover the HR pathway.31 Additionally, any protein alterations along these pathways can disrupt the dynamic equilibrium. CCCTC-binding factor (CTCF) enhances CtIP recruitment with its N-terminus and ZF domain as the binding site, thus improving the efficiency of HR, as well as alt-EJ and SSA when HR is suppressed.7 Moreover, CTCF can be modified by PARP1 in a process called PARylation. PARylized CTCF enables the recruitment of BRCA2, further allowing the loading of RAD51 to DSBs.32 Studies have revealed that the critical DNA pol θ in alt-EJ is often upregulated in cancer tissue but is absent in normal tissue. Pol θ can also bind to RAD51 and inhibit its nucleofilament formation, thus increasing the level to which pol θ can suppress HR.33
A total of 18 HDACs remove acetyl groups from histones and nonhistones, which are also called lysine deacetylases or KDACs. These members could be grouped into four types based on their structures. Class I HDACs (HDAC1, HDAC2, HDAC3 and HDAC8) are related to the yeast transcriptional regulator RPD3. Class II HDACs (HDAC4, HDAC5, HDAC6, HDAC7, HDAC9, and HDAC10) share high sequence homology with I1. Class III HDACs share an NAD+-binding catalytic domain. Finally, class IV members only include HDAC11, which is structurally related to both class I and II HDACs.34,35 Class I, II, and IV are referred to as ‘classical’ HDACs, whereas class III members are also named sirtuins (SIRTs, including SIRT1-7). Classical HDACs are Zn2+-dependent enzymes harboring a catalytic pocket with a Zn2+ ion at its base that can be inhibited by Zn2+-chelating compounds such as hydroxamic acids. Sirtuins (SIRTs) are derived from their homology Saccharomyces cerevisiae gene silent information regulation-2 (Sir2). SIRTs 1, 2, 6 and 7 are located in the nucleus, and SIRTs 1 and 2 can also be found in the cytoplasm. In the mitochondria, SIRTs 3, 4, and 5 can be found.36 HDACs not only epigenetically modify histone acetylation but also deacetylate various crucial factors associated with different biological processes, including the cell cycle, apoptosis, metabolism, immunity, and ROS production. Specially, an increasing number of studies have shown that HDAC inhibition-related histone acetylation decreases DNA repair and causes DNA damage that is significantly increased in solid tumors. Histone H2AX is a DNA damage sensor and is crucial for DNA integrity.37 Upon DNA damage, H2AX is phosphorylated at serine 139 to generate γH2AX. This phosphorylation event serves as an anchor for the accumulation of the signaling cascade initiated by DNA damage and requires the activation of DNA-PKcs, ATM, and ATR. Specifically, the acetylation status of H2AX on Lys5, which is regulated by TIP60 (a histone acetyltransferase) and SIRT1, plays an important role in the formation of γH2AX. The absence of SIRT1 leads to H2AX K5Ac hyperacetylation, lowering DDR levels.38 The acetylation of histone H4 mainly influences the choice of DNA repair pathway. In response to DNA damage, H4 acetylation follows a rise-fall pattern, which corresponds to rapidly occurring NHEJ and slowly occurring HR.39 TIP60 mainly contributes to the accumulation of BRCA1 (mediator of HR) and the inhibition of 53BP1 (mediator of NHEJ) at DSB chromatin, while HDACs play the opposite role.40 The acetylation of histone H3 regulated by HATs/HDACs is required for the binding of BRG1 to γH2AX nucleosomes, and SWI/SNF, γH2AX and H3 acetylation cooperatively act in a feedback activation loop to facilitate DSB repair.41,42 The regulation of DNA repair pathways by HDACs is summarized in Table 1 and Figure 6.39,43–96
HDAC1-3 are all highly expressed in HCC, correlating with tumor dedifferentiation and proliferative activity.97 HDAC1 and HDAC2 participate in the DDR through their location in DSB foci and coupling with accumulated γH2AX, regulation of H3K56 and H4K16 acetylation and requirement for DNA repair, particularly through NHEJ. Cells depleted of HDAC1 and HDAC2 showed DSB repair deficiency, while the DNA damage-induced phosphorylation of the checkpoint kinases CHK1 and CHK2 and the tumor suppressor p53 was higher and more sustained. Moreover, HDAC1/2 inhibition caused the NHEJ factors Ku70/80 and XRCC4 to show enhanced association with DSB sites.39 In addition to NHEJ, HDAC1, and HDAC2 also participate in the HR pathway through miR-182-related RAD51 regulation. Overexpression of miR-182 decreases RAD51, whereas HDAC1/2 can be recruited to the promoter of miR-182 to diminish its expression, thus promoting HR.44 In HCC tissue, miR-182 suppresses forkhead box protein (FOXO) 3a and activates the Wnt/β-catenin pathway, enhancing the progression and metastasis of tumor cells.45 TIP60 binds to H3K9me3 and transform it to H3K9ac, which acts as a symbol of active transcription, boosting tumor-related gene transcription, including cell cycle regulators, DNA damage-related genes and oncogenic genes. HDAC3 can attach to the H3K9ac site and reverse it to methylation, where DNA repair factors are allowed to initiate HR and NHEJ.46 Meanwhile, inactivation of HDAC3 also leads to the accrued acetylation of a series of sites on H4, such as H4K5/12/16, the accumulation of which precipitates a reduction in heterochromatin and genomic instability. In HCC, HDAC3 did not significantly increase or even decrease, whereas in HCC, with the loss of HDAC3, hepatoma-related pathways such as p53, g-glutamyltranspeptidase 1, and insulin-like growth factor II are all upregulated. 53BP1 and γH2AX also increase, indicating widely appearing DSB foci.47 Although there are few reports on the regulation of DNA repair by HDAC8, a test about therapy in acute myeloid leukemia with HDAC8 inhibitor reveals that several DNA sensors (pATM, CHEK1 and CHEK2) and DNA repair factors (CtIP, Rad51, and BRCA1 in HR; Ku70 and DNA-PKcs in NHEJ) are all markedly inhibited.49
HDAC4 is significantly upregulated in liver cancer and can remodel chromatin structure and control protein binding to DNA, thus regulating oncogenes. Knockdown or inhibition of HDAC4 reduces cell viability, the activation of AKT and the induction of apoptosis. RAD51 and γH2AX are decreased in HDAC4 knockdown HCC cells, indicating that HR repair can be regulated by HDAC4. Moreover, HDAC4 and Rad51 interact with the SUMO-conjugating enzyme Ubc9, and the HDAC4/Ubc9/RAD51 complex can act as a target of radiosensitization for DNA repair in HCC.53 In addition, HDAC4 can act as a SUMO E3 ligase. HDAC4 interacts with SUMOylation of SIRT1 to form the SIRT1-SUMO-1/HDAC4/Ubc9 complex and combines with hypermethylated in cancer 1 (HIC1), a tumor suppressor gene, to drive deacetylation and SUMOylation of HIC1. SUMOylated HIC1 then enhances its cooperation with MTA1, a component of the NuRD complex, to repress the transcription of target genes that favor the DNA repair process.50 HDAC7 has a repressive role by forming a complex with signal transducer and activator of transcription 3 (STAT3) and Tip60 to inhibit gene expression, which is related to STAT3-mediated transactivation. Tip60 binds with HDAC7 on its N-terminal zinc finger-containing region and is essential for its repressive function.54 Activated STAT3 further decreases the phosphorylation of checkpoint kinase 1 (CHK1), suppressing the intra-S phase cell cycle checkpoint activation. Phosphorylation deficiency of CHK1 impairs RAD51 nucleation, thus curtailing HR.56 Ku70 and scaffold matrix attachment region-binding protein 1 (SMAR1) aggregate at DSB sites. SMAR1 connects Ku70 and HDAC6 to form a triple complex to induce deacetylation of Ku70, promoting Ku70 binding to DSB sites.58 The combination of Ku70 and BCL2-associated X protein (BAX) depends on the deacetylation of Ku70 with HDAC6, the loss of which leads to the release of BAX, resulting in apoptosis.59 In high-grade serous ovarian carcinomas, HDAC6 removes acetylation from H4K12 and H4K16, inducing HR deficiency, which increases the sensitivity to chemotherapy.60 In contrast, HDAC6 activates its downstream factor, Sp1, to upregulate RAD51, CHEK1, EXO1, RAD54L, and GEN1, promoting HR repair in glioblastoma cells.61 As an epistatic gene of BRCA1, HDAC10 can compensate for the loss of BRCA1 in the cell repair process and reduce the appearance of DSBs. Although BRCA1 is lost, ovarian carcinoma cells can still exert their repair function by HDAC10, while loss of HDAC10 worsens the DSB repair defect. A study utilizing a tissue culture-based homology-directed repair assay revealed that depletion of HDAC9 or HDAC10 specifically inhibits the HR pathway in HeLa cells.57
The influence of SIRTs on cell viability can be attributed to their protection of telomeres and the activation of all SIRTs instead of only one SIRT, resulting in protection against metabolic disorders, age-related diseases and stem cell failure. The defensive function is based on NAD+ precursors, such as nicotinamide mononucleotide (NMN). During cell damage, the level of NAD+ is significantly decreased, which worsens telomere dysfunction. NMN helps to defend against liver fibrosis at the DNA level, which stabilizes telomeres together with SIRT1. In addition, SIRT6 can also combine with telomeres and deacetylate its H3K9 and H3K56 sites, which is essential for telomere capping. When telomeres are established, the repression of SIRTs is achieved by the DNA damage response and p53. During p53-dependent regulation, nonmitochondrial SIRTs are suppressed at the translational level, while mitochondrial SIRTs are transcriptionally regulated. The upstream factors of nonmitochondrial SIRTs are all highly selective. Nonmitochondrial SIRTs are affected by PGC-1α and PGC-1β, whereas mitochondrial SIRTs are regulated by miR-34a, 26a, and 145.98
SIRT1, the most thoroughly studied sirtuin, is involved in a host of biological behaviors in the liver, such as lipid metabolism, oxidative stress and inflammation. SIRT1 acts as a stress sensor and couples with cellular metabolic/energy status, regulating transcription factors such as ChREBP, SREBP-1c, PPARα, PGC-1α, NF-κB, WNT, FOXO family, p53, and p65. When confronted with damage triggers, SIRT1 deacetylates downstream proteins to preserve cell viability. Nonetheless, if extreme damage occurs, SIRT1 helps cells proceed through the apoptosis pathway. Some factors, such as alcohol consumption and a high-fat diet, can impair the function of SIRT1, leading to alcoholic and nonalcoholic fatty liver diseases.99,100 In liver tissue with ischemic injury, SIRT1 expression and activity are upregulated to compensate for injury. This function can be abrogated by SIRT1 knockdown.101 In HCC tissues, SIRT1 mainly acts as an oncogene that mediates tumorigenesis and chemoresistance, promoting HCC proliferation and indicating poor prognosis in patients with liver cancer.63 SIRT1 has been proven to participate in both the NHEJ and HR repair pathways via its nonhistone protein deacetylation function. Deacetylation of Ku70 blocks the migration of the proapoptotic factor BAX toward mitochondria, thus preventing mitochondrial apoptosis and giving rise to the NHEJ repair pathway with Ku70.65 SIRT1 is the most important deacetylase of Ku70, and inhibition of SIRT1 enhances Ku70 acetylation, thereby directly obstructing the NHEJ repair pathway.64 SIRT1 can also remove acetylation from KAP1, thus stabilizing the interaction between KAP1 and 53BP1 to respond to DSBs and promoting NHEJ.69 In addition, SIRT1 deacetylates HDAC1 and mediates the NHEJ repair function.70 Finally, SIRT1 is a crucial mediator of the SIRT1-FOXL2-XRCC5/6 axis. FOXL2, as a modulator between NHEJ and HR, can be deacetylated by SIRT1 on the lysine 124 residue to release XRCC5/6. This process is fulfilled by the recruitment of SIRT1 to the nucleus when DSBs occur. Freed XRCC5/6 constitutes the Ku complex to allow the NHEJ pathway and compete for HR.75 Inactivated SIRT1 causes a reduction in RAD51, indicating that the HR repair pathway is also regulated by SIR1.66 At DSB sites, SIRT1 recruitment depends on ATM, whereas ATM autophosphorylation is performed and stability is ensured by SIRT1, indicating a cooperative relationship between ATM and SIRT1. SIRT1 promotes HR by deacetylating important proteins, such as NBS1 and Rad51. However, high acetylation levels of NBS1 and Rad51 can conversely downregulate SIRT1 activation. In addition, acetylation on NBS1 can be substituted with phosphorylation by SIRT1, as well as ATM, to promote HR.67 Another mechanism by which SIRT1 regulates the HR repair pathway is via BRG1 deacetylation. BRG1 is one of the major components of the SWI/SNF complex and contributes to the cell cycle in HCC.73 PAR (activated PARP) recruits SIRT1 and BRG1 to DSB sites, where SIRT1 deacetylates BRG1 to release its ATPase activity to loosen the DNA structure, enhancing HR.72 SIRT1 also deacetylates nibrin and WRN helicase to promote MRN complex generation for HR initiation.68
As a negative regulator of stress, radiation-induced impairment can be attenuated by SIRT2 depletion-related DSB repair. Depletion of SIRT2 enhances the expression of several DNA repair proteins, including Rad51, Artemis, DNA ligase IV and XRCC4, therefore improving HR and NHEJ efficiency.76 SIRT2 deacetylates conserve lysine residues of BARD1 to enable BRCA1 binding, thus catalyzing BRCA1-BARD1 heterodimerization to maintain their mutual stability, promoting HR and prohibiting tumorigenesis.77 SIRT2 and SIRT3 are responsible for recruiting RAD51 to DSB sites and activating RAD52 by deacetylation, and the deacetylated RAD52 participates in RAD51 recruitment. Both RAD51 and RAD52 are responsible for initiating DSB end resection at the early stage of HR, thus maintaining genome integrity and stability.78 SIRT3 colocalizes with γH2AX and 53BP1. The recruitment of SIRT3 depends on ring finger protein 8 (RNF8), and SIRT3 removes acetylation from H3K56 and attracts 53BP1 to DSB sites to enhance NHEJ.80
SIRT6 acts as a longevity gene that wields various functions to retain cell viability in aging cells, such as maintaining genome integrity.81 In aging cells, SIRT6 and NHEJ are downregulated, while short-term calorie restriction is associated with increased levels of DNA-PK and SIRT6 to enhance NHEJ.89 As a DSB sensor for DDR initiation, SIRT6 is located in DSB sites to recruit repair proteins from HR and NHEJ and ATM to fulfill H2AX phosphorylation.82 SIRT6 possesses NAD+-dependent protein deacetylase activity and mono (ADP-ribosyl) transferase activity, acting as a cross point between DNA repair and transcription adjacent to DSB sites to guarantee successful DNA repair. SIRT6 mono-ADP-ribosylates stimulates PARP1 poly-ADP-ribose polymerase activity, which enhances HR repair factors such as Rad51, Rad51C, Rad52, and NBS1 under oxidative stress,83 SIRT6 translocates to the DNA damage site and displaces HP1 to CHD4 on H3K9. HP1 executes its tumor suppressive and homeostasis regulating function by targeting chromatin activation, which is characterized by H3K9me3. Loss of HP1 renders HCC cells more tolerant to cell stress and increases the possibility of transformation.90 SIRT6 cooperates with ATM and the chromatin remodeler CHD4 to promote chromatin relaxation and recruit the chromatin remodelers SNF2H and CtIP to the compacted chromatin in HR.89 The interaction between CHD4 and NuRD is associated with the deacetylation of HDACs and PARP. CHD4/NuRD plays an oncogenic role in EpCAM+ liver cancer stem cells and in HCC cells and promotes proliferation, migration, invasion, colony-forming ability and cell apoptosis by regulating histone epigenetic status and the DDR.91 The CHD4/NuRD complex also represses the expression of complements and downregulates the infiltration of CD8 T cells in HCC tissue.92 SIRT6 dislodges KDM2A from the chromatin by mono-ADP-ribosylation of the lysine demethylase JHDM1A/KDM2A, resulting in increased H3K36me2 levels. In liver cancer stem cells, KDM2A induces the demethylation of H3K36 in the promoter regions of the transcription factors such as NANOG, SOX4, and OCT4, leading to tumor progression.93 Furthermore, H3K36me2 promotes H3K9 trimethylation by HP1α binding, leading to the recruitment of RNA polymerase II and NHEJ factors to transiently suppress H3K9 trimethylation.86 SIRT6 binds with DNA-PK to form a macromolecular complex to activate the DNA-PK catalytic subunit to stabilize DNA-PKcs at chromatin adjacent to an induced site-specific DSB.87 SIRT6 can interact with Ku80 to enable Ku80 to combine with DNA-PKcs, enhancing DNA-PKcs phosphorylation for efficient NHEJ.88
SIRT7 can be mobilized to DSB sites to compact DNA during DNA end-joining. SIRT7 can be recruited by PARP1 to DSB sites to deacetylate H3K18Ac for 53BP1 loading to start NHEJ.94 Transient upregulation of Dicer releases overloaded SIRT7 from DSBs and prevents its recruitment to maintain the DNA open state, thus promoting NHEJ factors to DSBs to moderately enhance NHEJ.96 In ribosomal DNA (rDNA), it is necessary to maintain highly compact heterochromatin to inhibit HR between rDNA repeats and protect nucleolar architecture and genomic stability. SIRT7 recruits DNA methyltransferase 1 and SIRT1 to form heterochromatin and avoid HR, while a lack of SIRT7 leads to nucleolar fragmentation, rDNA, and genomic instability.95
Faultless DNA repair pathways such as HR and NHEJ in HCC render tumor cells viable after radiation and chemotherapy by stimulating the DNA damage response and avoiding apoptosis. DNA repair factors, such as DNA-PK, ATM, ATR, Ku70/80, and PARP1, contribute to repair progression and induce drug resistance and poor prognosis. Therefore, drugs targeting these factors have been administered in studies and clinical therapies. PARP inhibitors such as olaparib, niraparib and rucaparib have been proven to exert positive functions in patients with BRCA mutant ovarian cancers during phase I and II experiments.102 Meanwhile, inhibition of DSB recognition, end processing, and DNA ligation processes has been indicated to enhance radiation therapy efficiency. Amid HCC treatment, inhibitors such as olaparib have been proven to significantly reduce malignant tumor phenotypes, such as drug resistance and cancer stem cell survival. Seventeen PARPs constitute the PARP family, whose functions relate to DSB site recognition and the synthesis of poly (ADP-ribose). PARP1 is recognized as the most researched protein and has been shown to be directly related to HR and NHEJ. Similar to SIRTs, PARP1 exerts its function by relying on NAD+ substrates to synthesize PARs to target proteins such as PARP itself and other DNA repair factors.103 PARP1 is significantly upregulated in embryonic stem cells and residual liver tumors after sorafenib treatment but gradually decreases during hepatic differentiation, which is critical to HCC stem cell pluripotency, residual tumor survival, and the potential of HCC sorafenib treatment resistance.
HDACs participate in various cellular behaviors of tumorigenesis and are associated with apoptosis, proliferation, metastasis, and senescence of cancer cells by targeting various signaling pathways and DNA-binding sites. Thus, HDAC inhibitors present a promising clinical measure for the treatment of malignant carcinoma, and several HDAC inhibitors have already been approved for hematologic malignancies and lymphomas since 2006. Moreover, the success of the clinical trial of chidaniline in the treatment of hormone receptor-positive breast cancer brought new insight into HDAC inhibitors in solid tumor therapy, and now more than 20 clinical studies are ongoing for refractory, advanced and recurrent solid tumors, including HCC.104 HDAC inhibitors consist of four major types: hydroxamates, cyclic peptides, aliphatic acids, and benzamides.105 The pharmacological functions of the major listed HDAC inhibitors are summarized in Table 2.106–115 Vorinostat, the first FDA-approved HDAC inhibitor, was approved for the treatment of cutaneous T-cell lymphoma (CTCL). Vorinostat belongs to the hydroxamic acid class of inhibitors, and its targets include class I, II, and IV HDACs. Since the efficacy of vorinostat on CTCL was confirmed, many clinical trials were designed to develop it against advanced and refractory tumors, alone or in combination with other inhibitors. Vorinostat obstructs HCC proliferation and promotes apoptosis, similar to autophagy-induced cell death. It also induces NK-cell-dependent cytolysis by cell recognition and directly impedes DNA replication by blocking topoisomerase IIα.116 Moreover, vorinostat analogs acetylate histones and induce apoptosis by increasing the expression of tumor suppressor miRNAs.117 A phase I study of sorafenib and vorinostat in advanced HCC revealed that 13 of 16 patients had durable disease control. Although further study was terminated due to the high incidence of toxicities in patients, the efficacy of SAHA in the treatment of HCC deserves further exploration.118 Subsequent to vorinostat, the cyclic peptide romidepsin was approved by the FDA in 2009 to treat CTCL. Romidepsin is a natural compound that specifically inhibits HDAC1 and HDAC2. It was reported to inhibit HCC cells by activating both the Erk/cdc25C/cdc2/cyclin B pathway and the JNK/c-jnk/caspase-3 pathway, leading to G2/M phase arrest and cell apoptosis, respectively.119 Although there are no clinical trials about romidepsin in HCC therapy, several phase I/II studies of romidepsin alone or combined with other inhibitors to treat different solid tumors are ongoing (NCT01537744, NCT01638533, NCT01302808, NCT02393794, etc.). Valproic acid sodium (VPA), which mainly inhibits HDAC1, is a fatty acid with anticonvulsant properties that can be used to treat epilepsy. It disrupts the formation of single-strand-DNA-RFA nucleofilaments and the activation of ATR and CHK2 by suppressing the recruitment of RPA and ATR interacting protein (ATRIP) to DNA damage sites. VPA has been shown to promote HCC apoptosis by activating tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) and caspases 3/9.120 It also decreases cyclin A and D1 levels and increases p21 and p63 levels to block the cell cycle.121 Finally, VPA upregulates MHC class I chain-related molecules to avert tumor escape. Its pharmacologic effect mainly involves the regulation of malignant tumor cells, while it has little effect on normal tumors.120,122 Panobinostat, a novel broad-spectrum HDAC inhibitor, induces endoplasmic reticulum stress by activating caspases 4/12 and upregulates the autophagy-promoting factors Beclin1 and Map1LC3B, directly leading to apoptosis.123 Panobinostat also inhibits angiogenesis and metastasis by inhibiting the expression of N-cadherin, vimentin, TWIST1, VEGF and the gankyrin/STAT3/Akt pathway,124 At present, panobinostat has been approved for the treatment of multiple myeloma by both the US Food and Drug Administration and the European Medicines Agency. Interestingly, it is also expected to be used in HCC treatment, and two phase I clinical trials are ongoing (NCT00823290 and NCT00873002). Belinostat is a novel pan-HDAC inhibitor that has been developed in various solid tumors and hematologic malignancies. The US FDA granted accelerated approval for belinostat for the treatment of patients with relapsed or refractory peripheral T cell lymphoma (PTCL) due to its high rate of efficacy and low rate of adverse reactions. A phase I/II study of it for patients with unresectable HCC was completed in 2017, and the results showed that the mOS of patients reached 8.9 months and that the median PFS was 2.83 months (NCT00321594). Resminostat, a dose-dependent selective inhibitor of HDAC1/3/6, has a unique mechanism of action that may expand the therapeutic treatments available to patients with advanced HCC. A SHELTER study provided promising data that combined therapeutic approaches utilizing resminostat was useful for HCC patients who failed sorafenib. Resminostat may counterbalance or even reverse the resistance mechanisms to sorafenib and provide a survival benefit (NCT00943449).125
Although many HDACis have been developed, the vast majority of them have been proven to have no anticancer effect on solid tumors. In HCC, despite the well-described mechanisms of HDAC inhibition, no phase III clinical trial has been conducted to date. There are several reasons for the lack of phase III trials. First, many clinical trials of HDACi have shown various adverse effects, including bleeding, nausea, neurotoxicity, fatigue, vomiting, anemia, arrhythmia, myocardial hypertrophy, diarrhea, hypophosphatemia, and hyponatremia. Second, not all patients will have the same survival advantage with HDACi; therefore, predictive biomarkers of response and prognostic biomarkers of survival are necessary to design and accumulate patients best suited for clinical studies. Third, although HDACis play a positive role in improving patient survival and symptom control, in most cases, HCC cells develop drug resistance to HDACis, resulting in malignant phenotype regeneration and maintenance.
HCC, as one of the most commonly occurring malignant tumors in the world, is a high-profile medical issue, with an age-adjusted incidence of 10.1 per 100 000 person-years worldwide.1 The late stage of HCC allows little space for surgical treatment, giving great significance for chemotherapy. Overexpression of HDACs frequently occurs in HCC cells and crucially controls DNA repair and maintenance of the neoplastic phenotype. DNA repair protects cells from deadly DNA lesions. Some key repair factors that undergo acetylation are targets of HDACi. Dysregulation of DNA repair proteins by HDACis might explain the efficacy of HDACis in HCC therapy. An increasing number of HDACi are undergoing preclinical experiments and clinical trials against cancers. Despite the role of HDACi on pathways in other cell processes, such as cell proliferation and metastasis, which have been carefully studied, research on their influence on DNA repair pathways has not yet been carried out. As HDACs participate in DNA repair in multiple pathways, HDACi-mediated dysregulation of DNA repair combined with DNA-damaging chemotherapeutics may overload DNA repair machinery. New approaches using HDACi and DNA repair inhibitors in combination may overcome tumor progression to improve patient survival. |
|
PMC9647120 | Yourong Jiang,Siwei Zhang,Lu Tang,Rui Li,Jinglei Zhai,Suisui Luo,Yiman Peng,Xiaohang Chen,Lanlan Wei | Single-cell RNA sequencing reveals TCR+ macrophages in HPV-related head and neck squamous cell carcinoma | 27-10-2022 | human papillomavirus,head and neck squamous cell carcinoma,macrophages,single-cell sequencing,tumor microenvironment | The prognosis of human papillomavirus (HPV)-infected head and neck squamous cell carcinoma (HNSCC) is often better than that of HPV- cancer, which is possibly caused by the differences in their immune microenvironments. The contribution of macrophage, as a principal innate immune cell, to this phenomenon is still unclear. In this study, a single-cell atlas of 4,388 high-quality macrophages from 18 HPV- and 8 HPV+ HNSCC patients was constructed with single-cell RNA sequencing data. Eight macrophage subsets were identified from HNSCC, whereas their functional properties and developmental trajectory were delineated based on HPV status. Our results demonstrated that macrophages in HPV+ HNSCC exhibit stronger phagocytic ability, although the infiltration rate of macrophages decreased. From the results, a unique macrophage subset with TCR and CD3-specific signatures was identified from HPV-related HNSCC. These TCR+ macrophages potentially participate in the regulation of the TCR signaling pathway and phagocytosis. In conclusion, our results suggested that HPV could affect the infiltration rate, function, and differentiation of macrophages in HNSCC, whereas TCR+ macrophages play a critical role in the HNSCC microenvironment. These results provide new insights into the immune microenvironment of HNSCC and offer a valuable resource for the understanding of the immune landscape of HPV-related HNSCC, which will in turn help the development of immunotherapy strategies for the disease. | Single-cell RNA sequencing reveals TCR+ macrophages in HPV-related head and neck squamous cell carcinoma
The prognosis of human papillomavirus (HPV)-infected head and neck squamous cell carcinoma (HNSCC) is often better than that of HPV- cancer, which is possibly caused by the differences in their immune microenvironments. The contribution of macrophage, as a principal innate immune cell, to this phenomenon is still unclear. In this study, a single-cell atlas of 4,388 high-quality macrophages from 18 HPV- and 8 HPV+ HNSCC patients was constructed with single-cell RNA sequencing data. Eight macrophage subsets were identified from HNSCC, whereas their functional properties and developmental trajectory were delineated based on HPV status. Our results demonstrated that macrophages in HPV+ HNSCC exhibit stronger phagocytic ability, although the infiltration rate of macrophages decreased. From the results, a unique macrophage subset with TCR and CD3-specific signatures was identified from HPV-related HNSCC. These TCR+ macrophages potentially participate in the regulation of the TCR signaling pathway and phagocytosis. In conclusion, our results suggested that HPV could affect the infiltration rate, function, and differentiation of macrophages in HNSCC, whereas TCR+ macrophages play a critical role in the HNSCC microenvironment. These results provide new insights into the immune microenvironment of HNSCC and offer a valuable resource for the understanding of the immune landscape of HPV-related HNSCC, which will in turn help the development of immunotherapy strategies for the disease.
Head and neck squamous cell carcinoma is the sixth most common cancer worldwide with an annual incidence of 650,000 cases and arises from the mucosal epithelium in the oral cavity, oropharynx, nasopharynx, and larynx (1, 2). The risk factors of HNSCC include smoking and alcohol consumption, whereas human papillomavirus has emerged as a novel pathogenic factor (1, 2). Clinically, the prognosis of HPV+ HNSCC patients is better, probably because of the impact of HPV on the immune cells in the tumor microenvironment (TME) (3). However, the mechanism underlying this difference in prognosis remains unclear. Macrophages represent a major constituent of immune cells in the TME and are able to stimulate key steps in tumor progression. They usually exhibit remarkable plasticity after being recruited into the TME and are skewed away from the tumoricidal phenotype (M1) toward a tumor-promoting phenotype (M2) (4, 5). For HNSCC, most infiltrated macrophages are M2-like tumor-associated macrophages (TAMs) which are correlated with tumor metastasis and poor prognosis (6, 7). Compared with HPV- HNSCC, a higher M1/M2 ratio of infiltrating macrophages in HPV+HNSCC may be associated with its favorable prognosis (8). However, the specific function of macrophages in HNSCC has not been fully described. It is necessary to systematically characterize macrophages within the HNSCC microenvironment. Recent developments in single-cell RNA sequencing (scRNA-seq) have enabled the classification of macrophages. In recent years, a unique subset of macrophages, TCR+ macrophages, has been discovered to exist in substantial tumors and diseases (9). TCR+ macrophages have been reported to be present in tuberculous granulomas, atherosclerosis (10), and several types of cancer (colon cancer, esophageal cancer, hepatic carcinoma, and melanoma) (10, 11); these unique cells could enhance phagocytosis and secrete IFN-γ, TNF, MIP-1β, and CCL2 to promote inflammation (11–13). However, the presence and functions of these cells in HPV-related HNSCC are still unknown. Here, we analyzed and characterized macrophages at single-cell resolution using a systematic approach. To the best of our knowledge, this is the first study to provide a comprehensive analysis of macrophage subsets and the effect of HPV infection on the regulation of macrophage subsets in HNSCC. Moreover, our analyses demonstrate the presence of TCR+ macrophages which are involved in the TCR signaling pathway and exhibit a stronger phagocytic ability in the HNSCC TME for the first time. Altogether, these analyses provide new insights into the immune microenvironment of HNSCC and could serve as a valuable theoretical basis for future studies on the immune regulation of HPV-related HNSCC.
HNSCC samples were collected after tumor resection surgery of patients with head and neck squamous cell carcinoma at the Cancer Hospital Chinese Academy of Medical Sciences, Shenzhen Center. All participants provided written informed consent regarding this study, and ethical approval for the study was obtained from the Ethics Committee of The Third People’s Hospital of Shenzhen [2021-057].
The scRNA-seq data of 18 HPV- HNSCC and eight HPV+ HNSCC patients used in this article were obtained from the GEO database (https://www.ncbi.nlm.nih.gov/geo/), and the accession number was GSE139324. RNA-seq data and clinical information of 520 HNSCC patients were obtained from TCGA database (https://www.cancer.gov/).
After downloading scRNA-seq data from the GEO dataset, we first processed the data with Seurat 3.0 in R (version 4.1.2) to remove low-quality cells. Cells would be flagged as poor-quality ones if they met one of the following thresholds: 1) the number of expressed genes lower than 500 or larger than 3,500; 2) 15% or more of UMIs were mapped to the mitochondria; 3) 0.1% or more of UMIs were mapped to hemoglobin genes; and 4) 3% or more of UMIs were mapped to ribosomal genes. Mitochondrial, ribosomal, and hemoglobin genes were filtered to avoid interference with subsequent analysis. A total of 20,326 genes in a total of 58,656 cells were detected. Then, we utilized functions in the Seurat package to normalize and scale the single-cell gene expression data. Then, principal component analysis (PCA) was performed according to the standard analysis process and identified 28 significant principal components. Cell clustering analysis was carried out with the parameter resolution of 1.2 and 31 Seurat clusters found. These clusters were annotated with marker genes from CellMarker (http://biobigdata.hrbmu.edu.cn/CellMarker/) (14). Macrophages and monocyte-derived macrophages were extracted while dimension reduction was performed using PCA with 19 significant principal components. These cells were then clustered with the parameter resolution of 0.6, and 11 clusters were identified. Among these 11 clusters, three of them were removed from subsequent analysis as one cluster was identified as epithelial cells and two clusters contained too few cells to be analyzed.
Pseudotime trajectory analysis was constructed based on Monocle2 (version 2.18.0, R package) following the tutorial to order the macrophage subsets in HPV+/- HNSCC. In order to avoid omitting some important genes, we selected all genes for completely unsupervised trajectory analysis. For differentiation trajectory analysis, top 50 pseudotime-related differentially expressed genes were selected and divided into three clusters for further analysis.
Cell–cell communication atlases between different cell subsets were explored with CellChat (Version 1.1.3, R package) following the standard protocol (15).
The top 50 genes (or all genes in the case of less than 50) with either significant upregulation or downregulation were selected based on results of differential gene expression analysis. Functional enrichment of selected genes was performed using DAVID 6.8 (https://david.ncifcrf.gov/) (16). A pathway enrichment analysis of different immune cells was performed with irGSEA (R package, https://github.com/chuiqin/irGSEA/). Phagocytosis and antigen presentation function of macrophage subclusters were analyzed by GSVA (R package,1.38.2). All genes associated with antigen presentation and phagocytosis available in Molecular Signatures Database (MSigDB, v7.5.1) were collected and used as reference genes for pathway enrichment analysis.
Immune infiltration of eight macrophage subsets was analyzed by CIBERSORT. We categorized the samples into high or low infiltration groups based on the mean infiltration rates of immune cells of the patients. Survfit (version 3.3-1, R package) and Survdiff (version 0.4.9, R package) were used to generate Kaplan–Meier survival curves and calculate the p-value of the log-rank test. Survival analysis of total macrophage infiltration in HNSCC was calculated by Timer 2.0 (https://cistrome.org) (17).
Immunofluorescence staining was performed as previously reported (18). Briefly, paraffin-embedded (FFPE) samples were sectioned at 3-µm thickness. A 0.01-M citrate buffer solution (pH 6.0) was used for retrieval treatment for 15 min in a pressure cooker. After blocking with 5% BSA at room temperature for 30 min, a mixture of three different primary antibodies, namely, CD3 (1:200 dilution, Cat No.17617-1-AP, Proteintech), CD68 (1:100 dilution, Cat No. 66231-2-lg, Proteintech), and TCR α (1:100 dilution, Abcam, ab18861), was added and incubated overnight at 4°C. Then, slides were incubated with the mixture of Coralite488-conjugated anti-mouse IgG (H + L) secondary antibody (1:250 dilution, Cat No. SA00013-1, Proteintech) and Coralite594-conjugated anti-rabbit IgG (H + L) secondary antibody (1:250 dilution, Cat No. SA00013-4, Proteintech) at 37°C for 30 min. A fluorescent quenching sealing agent with DAPI was dropped on the slides and covered with slip carefully. Images were taken using a Leica TCS Sp8 laser scanning confocal microscope at the appropriate excitation wavelength for the fluoroscope.
The statistical significance for differential gene analysis of scRNA-seq data was calculated by Wilcox test using the Seurat R package. The statistical significance in the proportions of different infiltrated immune cells and the infiltration rate of macrophage subsets among the immune cells in HPV+/- HNSCC samples were assessed using the Mann–Whitney test in GraphPad Prism 6 software. Statistical significance was determined at p < 0.05. All experiments were repeated three times and at least three samples per experiment.
To explore the characteristics of macrophages in HPV+/- HNSCC, scRNA-seq data of tumor-infiltrating immune cells from 18 HPV- and 8 HPV+ HNSCC patients were obtained from the GEO dataset (GSE139324) with a total of 58,656 quality control compliant cells. Dimension reduction and unsupervised cell clustering were performed using the Seurat R package. B cells (CD79A), plasma cells (CD79A and IGHG1), T cells (CD3E), NK cells (KLRD1), myeloid cells (CD14, CD68, and LYZ), mast cells (TPSAB1), and plasmacytoid dendritic cells (LILRA4) were firstly identified from the t-distributed stochastic neighbor embedding (t-SNE) result ( Figure 1A ), and the representative characteristic gene expressions of each cluster were shown ( Figure 1C ; Supplementary Figure 1A ). The proportion of infiltration of different cell clusters in each patient is illustrated in Figure 1B . As compared with HPV- HNSCC patients, the proportion of B cells increased while the proportions of myeloid cells and NK cells decreased in HPV+ HNSCC ( Figures 1B, D ). Based on the expression of the feature markers, myeloid cells were further classified into macrophages (APOE, CD163, and FCGR3A), monocyte-derived macrophages (FCN1 and FCGR3A), classical dendritic cells (CD1C and FSCN1), and monocytes (FCN1) ( Figure 1E ). The feature markers of these clusters are shown in Figures 1H, I . The proportion of these four clusters in each patient was calculated. The results showed that the infiltration rate of macrophages and monocyte-derived macrophages increased significantly in HPV- HNSCC ( Figures 1F, G ). As calculated by Timer 2.0, lower macrophage infiltration in HPV+ HNSCC was positively correlated with the better prognosis of 98 patients from TCGA dataset ( Figure 1J ). Moreover, the phagocytosis score of macrophages was significantly increased in HPV+ HNSCC ( Figure 1K ). These results suggested that HPV infection may affect the infiltration level and function of macrophages in HNSCC.
Based on the scRNA-seq data, we categorized the macrophages into eight subclusters and observed a high heterogeneity among the clusters ( Figure 2A ). The gene expression and enriched pathways of these eight macrophage subclusters were also different ( Figures 2C, E ). A functional enrichment of the significantly upregulated genes in these eight subclusters was analyzed and is listed in Supplementary Table 1 . In addition, the significantly enriched biological processes in macrophage subsets were compared between HPV+/- HNSCC. The results showed that macrophages in HPV+ HNSCC were involved in DNA damage repair pathways ( Supplementary Figures 1C, D ), which is consistent with our previous findings (19). Next, we compared the levels of infiltration of macrophage subsets between HPV+/- HNSCC patients and found that infiltration rates of C0, C1, and C2 cells were lower in HPV+ HNSCC ( Figures 2B, D ), whereas there was no significant difference in that of the other subsets. Then, the effect of HPV on the gene expression of these three groups was analyzed, respectively. The functional enrichment of the highly differentially expressed genes in HPV+ HNSCC patients were analyzed, and the results showed that the macrophage-related immune response was more active ( Supplementary Figure 1B ). Remarkably, the antigen presentation and phagocytosis of macrophage-related genes, HLA-DQA2 and IGLC2, were highly expressed in all these three macrophage subsets in HPV+ HNSCC patients ( Supplementary Figure 2 ). Moreover, the patients’ survival rates assessed using TCGA data showed that the prognosis of patients with higher C1 and C2 infiltrations was better ( Figure 2F ), whereas there was no significant difference in that of HPV- HNSCC patients. These results suggested that HPV could simultaneously affect the infiltration level and function of macrophage subsets and subsequently alter the prognosis of HNSCC.
Next, the differentiation trajectories of macrophages were analyzed under different HPV statuses by pseudo-temporal analysis in order to determine whether HPV would affect the differentiation of macrophages. The results showed that the trajectories of macrophage cell subsets in HPV+/- HNSCC were roughly the same ( Figures 3A, C ). In addition, the differential gene expression along trajectories were analyzed and the top 50 genes with the most significant difference were selected for downstream analysis. These genes can be clustered into three groups, according to their changes in the trajectories. The composition of the genes in these three clusters between HPV+/- HNSCC varied slightly, suggesting that the hub genes involved in cell differentiation were different between HPV+/- HNSCC ( Figures 3B, D ). The functional enrichment analysis of these genes was carried out, and the results showed that the antigen presentation ability of macrophages significantly increased in HPV+ HNSCC ( Figures 3E, F ). These results showed that HPV had functional and quantitative effects on infiltrated macrophages around HNSCC.
Interestingly, we found a C2-specific subset of macrophages expressing TCR marker genes (TRAC and TRBC1) and CD3-specific signatures (CD3E) ( Figures 4A, B ). Totally, TCR+ macrophages account for 0% to 46% of all macrophages across patients ( Supplementary Figure 3 ). In addition, immunofluorescence analysis of HNSCC samples showed the co-localization of CD68 and CD3/TCRα on the cell membrane, respectively ( Figure 4C ; Supplementary Figure 3A ), and these cells were usually isolated in areas with high concentrations of T cells, suggesting the association between this group of cells and T cells. In order to verify the characteristics of TCR+ macrophages, the single-cell sequencing data of T cells and macrophages were analyzed. The results showed that the characteristics of TCR+ macrophages were more consistent with the characteristics of macrophages, but the gene expressions of TCR and CD3 were higher than those of ordinary macrophages ( Supplementary Figures 4A–C ). These results showed that TCR+ macrophages were a special group of macrophages rather than T cells.
To further explore the potential role of TCR macrophages in HNSCC, the differentially expressed genes between TCR+/- macrophages were compared ( Figure 5A ). Among them, essential genes for TCR signaling essential genes including LCK, FYN, LAT, and ITK as well as GZMA and GZMM involved in cytotoxic effects were significantly increased in TCR+ macrophages ( Figure 5B ). In addition, the results of the GO analysis showed that upregulated genes of TCR+ macrophages were enriched in TCR signaling pathways, T-cell stimulation, activation, and differentiation pathway and cytolysis pathway, whereas the downregulated genes were enriched in positive regulation of NF-κB activity and IL-4 ( Figure 5C ). Furthermore, we evaluated the cellular interactions between TCR+ macrophages and other cells in HPV+/- HNSCC. The results showed that TCR+ macrophages have differential cellular interactions with other immune cells compared with TCR- macrophages ( Figure 5D ). The cellular communication network of TCR+ macrophages was also different in HPV+/- HNSCC ( Figure 5E ). The phagocytic ability of TCR+ macrophages was stronger than that of TCR- macrophages in HNSCC ( Figure 5F ), but there was no significant difference between HPV+/- HNSCC ( Figure 5H ). However, the upregulated genes in TCR+ macrophages could participate in complement-related pathways in HPV+ HNSCC ( Supplementary Figure 1B ). The TCR+ macrophage was associated with the myc target v1 pathway in HPV-HNSCC but not in HPV+HNSCC ( Supplementary Figures 1C, D ). These results suggested that HPV infection influences the frequency of infiltration and specific cellular functions of TCR+ macrophages.
ScRNA-seq data provided a comprehensive resource for understanding the characteristics of macrophage subsets. A previous study has analyzed the unique states and potential plasticity of myeloid cells in HPV-related HNSCC (20). To the best of our knowledge, this is the first study providing a deep insight into the single-cell atlas of macrophage subsets and the regulation of HPV on them in HPV-related HNSCC. In addition, we identified TCR+ macrophages and its functional properties in HPV-related HNSCC for the first time. These results offer a valuable resource for understanding the immune landscape and immunotherapy strategies for HPV-related HNSCC. The involvement of HPV infection in the HNSCC microenvironment always leads to a higher degree of immune response which is related to a significantly favorable prognosis (21). The infiltration percentage of the prognostic factor, macrophage, decreased significantly in HPV+ HNSCC based on the results of the scRNA-seq analysis in our study, while Kürten et al. also found the same phenomenon (22). Macrophages are multifunctional plastic cells and are easily modulated by factors in the microenvironment which could promote them to polarize into M1 (antitumor type) or M2 type (pro-tumor type) (23). During tumor progression, macrophages usually transform into M2 type eventually, which may be related to a favorable prognosis of HPV+ HNSCC (24). In breast cancer, gastric cancer, rectal cancer, and pancreatic cancer, survival analysis showed that low levels of M2 macrophage infiltration were associated with a better prognosis (23). Previous studies have shown that HPV+ and HPV- HNSCC cell lines could induce macrophage polarization into M1 and M2 types, respectively. The ratio of M1/M2 infiltration was significantly increased in HPV+ HNSCC, which contributed to a favorable prognosis (8). Barbora et al. also discovered that in comparison to HPV+ tumors, HPV- HNSCC was more infiltrated by M2 TAMs (23). HPV-promoted tumor cells secrete IL-6, thereby increasing radiosensitivity through M1 polarization of macrophages (25). Another study proved that HPV+ HNSCC-derived exosomal miR-9 induces macrophage polarization to the M1 type (26). In this study, we observed that macrophages in HPV+ HNSCC have the same dynamic change. According to the result of pseudotime analysis, the signature gene expressions of M1 macrophages in HPV+ HNSCC decreased more slowly, which stresses the importance of HPV’s effect on the macrophage differentiation ( Supplementary Figure 5A ). The phagocytotic ability was enhanced in HPV+ HNSCC, reconfirming that HPV can affect the function of macrophages. In our study, we found that lower macrophage infiltration in HPV+ HNSCC was positively correlated with better prognosis, but the phagocytosis score of macrophages was significantly increased in HPV+ HNSCC. This is possibly because most of the infiltrates around HNSCC are M2-type macrophages. The presence of HPV reduces the M2 macrophages in the tumor microenvironment and promotes the polarization of macrophages to M1 (26), so the phagocytosis of macrophages is increased. In this study, we have classified macrophages into eight populations with different genetic characteristics and functions. Most studies classify macrophages into M1 and M2, but this classification is not applicable for single-cell analysis ( Supplementary Figure 5B ). Clément et al. tried to define aortic macrophage heterogeneity into resident-like macrophages, inflammatory macrophages, and TREM2hi macrophages at the single-cell level (27). Our data could find similar macrophage subsets ( Supplementary Figure 5C ), but in fact, there are still differences among the three cell subsets. Furthermore, although both C0 and C1 have the same antigen presentation function, the expression of C1QA and APOE in the C0 cell group was much lower than that in the C1 group, and the C1 group mainly belonged to monocyte-derived macrophages ( Supplementary Figure 6 ). Therefore, we evaluated the resolution parameter in the analysis with Seurat and retained the original subgroup classification to facilitate a better comparison of the characteristics of macrophage subsets. However, as previously discussed, there are still some limitations in analyzing macrophages with biological information, which need algorithm optimization and experimental verification in the future. The presence of TCR on macrophages is unconventional; however, recent studies have reported TCR expression in macrophages (9). These special macrophages also express CD3 and other molecules that are necessary for TCR signaling. In this study, the expression of essential genes for TCR signaling including LCK, FYN, LAT, and ITK was found to be significantly increased in TCR+ macrophages ( Figure 5B ). In addition to HNSCC, TCR+ macrophages have been reported to be present in tuberculous granulomas, atherosclerosis (10), and several types of cancer (colon cancer, esophageal cancer, hepatic carcinoma, triple-negative breast cancer, and melanoma) (10, 11, 28). Adriana et al. have proved that human circulating monocytes could be differentiated into CD3+TCR+ macrophages in tuberculosis (29). Zhang et al. proved that Japanese encephalitis virus infection could induce the differentiation of myeloid-derived suppressor cells into CD3+ macrophages in the brain (13). In this study, scRNA-seq data analysis also revealed that this population of cells may have been derived from monocytes ( Supplementary Figure 6 ), but further experimental verification is needed. Several studies have shown that these unique cells could enhance phagocytosis and secrete IFN-γ, TNF, MIP-1β, and CCL2 to promote inflammation (11–13). Our results here indicated that the enhanced phagocytic ability of TCR+ macrophages may be critical to the prognosis of HNSCC patients. However, the inactivation of immune response in HPV- HNSCC may limit the function of TCR+ macrophages. Therefore, despite of the increased number of TCR+ macrophages, the amount of TCR+ macrophages were not sufficient to alter the prognosis of HPV- HNSCC patients. There are still a lot of more studies to be done to thoroughly understand the functions of TCR+CD3+ macrophages. In summary, our transcriptional map of macrophages from HNSCC provided a framework for understanding the function of macrophages and revealed the dynamic nature of macrophages in HNSCC. In addition, learning the functions of TCR+ macrophages from many aspects will help us to understand the immune landscape in HNSCC patients and serves as an essential resource for further exploration of the roles of macrophages in HNSCC and other tumor types.
The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.
The studies involving human participants were reviewed and approved by The Third People’s Hospital of Shenzhen. The patients/participants provided their written informed consent to participate in this study.
YJ, SZ, LT, RL, JZ, SL, YP, XC, and LW contributed to the conception and design of the study. YJ, SZ, and LT performed the experiments and data analyses. YJ and SZ wrote the first draft of the manuscript. All authors contributed to the manuscript revision and read and approved the submitted version.
The study was sponsored by the National Science Fund for Distinguished Young Scholars (No. 82102822), the Shenzhen Science and Technology Innovation Program (JCYJ20210324131607019), the Shenzhen Science and Technology Innovation Program (KCXFZ20211020163544002), and the Science and Technology Planning Projects Shenzhen Municipality (JCYJ20210324111810028).
This research was supported by the Center for Computational Science and Engineering at Southern University of Science and Technology. We would like to thank Off-campus Practice Teaching Base of Biomedical Technology (Engineering) of Southern University of Science and Technology. We are grateful to the Cancer Hospital Chinese Academy of Medical Sciences, Shenzhen Center, for providing technical guidance and HNSCC patient tissues. We also give thanks to the Wu Lien-Teh Institute and the Key Laboratory of Preservation of Human Genetic Resources and Disease Control in Harbin Medical University for providing the experimental platform.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647127 | Srishti Chakravorty,Behdad Afzali,Majid Kazemian | EBV-associated diseases: Current therapeutics and emerging technologies | 27-10-2022 | high-throughput sequencing technologies,EBV therapeutics,EBV animal models,molecular mechanisms of EBV-host interactions,EBV-associated diseases and cancers,EBV vaccines | EBV is a prevalent virus, infecting >90% of the world’s population. This is an oncogenic virus that causes ~200,000 cancer-related deaths annually. It is, in addition, a significant contributor to the burden of autoimmune diseases. Thus, EBV represents a significant public health burden. Upon infection, EBV remains dormant in host cells for long periods of time. However, the presence or episodic reactivation of the virus increases the risk of transforming healthy cells to malignant cells that routinely escape host immune surveillance or of producing pathogenic autoantibodies. Cancers caused by EBV display distinct molecular behaviors compared to those of the same tissue type that are not caused by EBV, presenting opportunities for targeted treatments. Despite some encouraging results from exploration of vaccines, antiviral agents and immune- and cell-based treatments, the efficacy and safety of most therapeutics remain unclear. Here, we provide an up-to-date review focusing on underlying immune and environmental mechanisms, current therapeutics and vaccines, animal models and emerging technologies to study EBV-associated diseases that may help provide insights for the development of novel effective treatments. | EBV-associated diseases: Current therapeutics and emerging technologies
EBV is a prevalent virus, infecting >90% of the world’s population. This is an oncogenic virus that causes ~200,000 cancer-related deaths annually. It is, in addition, a significant contributor to the burden of autoimmune diseases. Thus, EBV represents a significant public health burden. Upon infection, EBV remains dormant in host cells for long periods of time. However, the presence or episodic reactivation of the virus increases the risk of transforming healthy cells to malignant cells that routinely escape host immune surveillance or of producing pathogenic autoantibodies. Cancers caused by EBV display distinct molecular behaviors compared to those of the same tissue type that are not caused by EBV, presenting opportunities for targeted treatments. Despite some encouraging results from exploration of vaccines, antiviral agents and immune- and cell-based treatments, the efficacy and safety of most therapeutics remain unclear. Here, we provide an up-to-date review focusing on underlying immune and environmental mechanisms, current therapeutics and vaccines, animal models and emerging technologies to study EBV-associated diseases that may help provide insights for the development of novel effective treatments.
Oncogenic viruses cause approximately 15-20% of all human cancers (1, 2). According to the International Agency for Research on Cancer (IARC), there are seven major human oncogenic viruses (3). These include DNA viruses, such as Epstein–Barr virus (EBV; also known as HHV4), Kaposi sarcoma-associated herpesvirus (KSHV; also known as HHV8), Hepatitis B virus (HBV), human papillomaviruses (HPV) and Merkel cell polyomavirus (MCPyV) and RNA viruses, such as human T-lymphotropic virus 1 (HTLV-1) and Hepatitis C virus (HCV). Despite many differences, these viruses have evolved common mechanisms to persist and replicate within host cells and facilitate escape of infected cells from the host’s immune surveillance. EBV and KSHV are two oncogenic viral agents of the γ-Herpesviridae subfamily that are known to modulate a plethora of biological processes in viral-associated cancers. The γ-Herpesviridae family is divided into two genera: Lymphocrytoviridae which includes EBV and Rhadinoviridae which includes KSHV. γ-herpesviruses encompass a broad range of pathogens in lower mammals ranging from murine herpesvirus-68, bovine herpesvirus 4 and equine herpesvirus 2 that closely resemble the rhadinovirus. Interestingly, to date, lymphocryptoviruses have been found only in primates and humans (4). EBV was first discovered by Epstein, Achong, and Barr in 1964 who isolated this virus from the cells of a Burkitt lymphoma (BL) patient in Africa (5, 6). Since then, it has become evident that EBV infects ~95% of the world’s adult population. The typical transmission route is through bodily fluids, such as saliva, where the orally transmitted virions infect resting B and epithelial cells of the oral cavity. Primary infection is typically asymptomatic, although 35-50% of the human adolescent population develop infectious mononucleosis (IM) approximately 1 month after infection, and the virus persist throughout an individual’s life (7–10). After acute infection, a dormant state is established due to a strong, virus-specific T cell response (7). However, when the balance between the virus and host immune system is disrupted, EBV can drive malignant transformation of both lymphoid and epithelial origins, causing ~200,000 deaths annually (11–13). As a herpesviruses, EBV can cause either latent or lytic infection. In epithelial cells, EBV typically undergoes lytic replication. In B cells, EBV usually establishes lifelong latency with rare sporadic reactivations. During latency only a few essential viral genes are expressed and production of virions are stalled (14–17). The switch from latent to lytic phase is governed by several factors (18, 19). While EBV-encoded products in both phases can play a role in transformation and tumorigenesis, the literature is more extensive on the oncogenic role of latent genes compared to lytic genes. However, it is challenging to target latent EBV using current immunotherapeutic strategies, specifically due to reduced antigen expression. As a result, patients with EBV+ or EBV– tumors are typically subjected to similar treatment regimen. This underscores the need to investigate the complexity of EBV-host interactions to help the development of EBV-specific cancer therapies. In this review, we will first discuss the EBV lifecycle and different types of EBV-associated malignancies. We will then summarize the major underlying molecular mechanisms and therapeutic strategies for EBV+ cancers. Lastly, we will discuss some of the preclinical animal models and emerging technologies for investigating different aspects of host-pathogen interactions in EBV-associated malignancies.
Epstein-Barr Virus (EBV) exhibits a biphasic lifecycle that includes latent and lytic (replicative) phases (20). Upon infection the virus typically establishes latency within the host cell (21). During latency, only a handful of latent genes that are necessary for the maintenance and persistence of the viral genome are expressed. EBV encodes eight latency genes whose expression in host cells and/or malignancies defines EBV latency programs (20). Based on which of the eight latent viral genes are expressed, viral infection is categorized into three main latency programs, latency III, II and I/0 (22). EBV-infected naïve B cells exhibit a latency III program, which allows for the proliferation and expansion of infected cells (23). Latency III genes include 6 EBV nuclear antigens (EBNA1, 2, 3A, 3B, 3C, LP), 2 latent membrane protein (LMP1 and LMP2), EBV-encoded small RNAs (EBERs), and EBV-encoded microRNAs (miRNAs) (24, 25). The cells in this latency program are highly immunogenic and can be rapidly eliminated by the host immune response, specifically by EBV-specific T cells (26). Latency II has a more restricted expression of EBV genes, namely EBNA1, LMP1, and LMP2A/B making them less immunogenic. Eventually, EBV sequentially shuts down the expression of all the latent genes except EBNA1 and a few EBV-encoded RNAs in latency I. Latency II can also be divided to IIa and IIb based on the expression of LMPs and EBNA2-3 (IIb is EBNA2-3+LMP–; IIa is EBNA2-3–LMP+). In most individuals, EBV persists quiescently within a subset of memory B cells (<0.005% B cells in the peripheral blood) without expressing any viral genes in latency 0 state, also referred to as a ‘true latency’ (24, 27, 28). Latent EBV genes are reported to promote tumorigenesis, inhibit apoptosis, and suppress recognition of infected cells by host immune cells (29). EBV-related malignancies are linked with different EBV latency programs. Lymphoproliferative disorders that are commonly associated with immunosuppression such as post-transplant lymphoproliferative diseases (PTLDs) and acquired immunodeficiency syndrome (AIDs) associated lymphomas exhibit latency III (29). Hodgkin lymphoma, T/NK cell lymphomas and nasopharyngeal carcinoma (NPC) exhibit latency II (30). Gastric carcinoma and Burkitt lymphoma exhibit latency I program (27, 31). So far, EBV in latency 0 has not been associated with any malignancies, presumably due to dormancy during this program ( Figure 1 ). The lytic phase is necessary for EBV progeny production and horizontal transmission of virus from host to host, so represents an integral aspect of viral pathogenesis (32). The switch from latent to lytic cycle can be either spontaneous or chemically induced. Some of the commonly used agents to induce lytic cycle include phorbol esters (PMA), sodium butyrate, calcium ionophores, DNA methyltransferase inhibitors (DNMTi), transforming growth factor-beta (TGF-β), doxorubicin and gemcitabine (because these are stress inducing chemotherapeutic drugs) and anti-IgG or anti-IgM as B-cell receptor stimulants (33–35). During lytic reactivation, the full repertoire of >80 viral genes is temporally regulated and expressed during three phases - immediate early (IE), early (E), and late (L). The first phase is primarily initiated by BZLF1 (ZEBRA) and BRLF1, the two key EBV immediate-early (IE) lytic transcription factors. Both genes function to promote their own and each other’s expression, as well as the expression of viral E genes, that code for proteins needed for viral replication (e.g., viral DNA polymerase). BZLF1 forms a homodimer via its basic leucine zipper motif and binds to BZLF1-responsive elements (ZRE) on DNA (36). The binding of BZLF1 to CpG methylated DNA leads to activation of several lytic viral genes that are silenced in latent cells by CpG methylation (37, 38). In addition, binding of BZLF1 to the origin of lytic replication (oriLyt) ZRE promotes lytic viral DNA synthesis (39). Similarly, BRLF1 binds to the BRLF1-responsive elements (RRE) on DNA and is reported to induce lytic replication via the PI3K and ERK signaling pathways (40, 41). Both BZLF1 and BRLFI are quintessential for EBV lytic replication since knocking out these genes blocks the latent to lytic switch. In addition, overexpression of BZLF1 and BRLF1 in latently infected cells can induce EBV lytic reactivation (42). This lytic induction leads to a cascade of viral gene expression, which promote viral DNA replication and virion production. Following viral replication, late viral genes code for structural proteins, such as gp350/220 encoded by the BLLF1 gene are expressed (32). Interestingly, during lytic DNA replication in γ-herpesviruses, continuous DNA synthesis is needed for the transcription of late lytic viral genes but not for early lytic genes (20, 43). The virions can disseminate viral particles within host cells and among hosts. EBV replicates in latency I, II and III via proliferation of activated B cells. Interestingly, lytic replication can only be efficiently induced from latency I/0, and after extensive methylation of the viral genome. This is because BZLF1 prefers binding to methylated CpG sequences to initiate infectious particle production (38, 42).
The EBV genome is packaged similarly to that of host cells, that is to say into nucleosomes, except loci that harbor the origin of plasmid replication (OriP) (44, 45). Nucleosome folding along with several histone modifications at the promoter of lytic genes keep them transcriptionally silent during latent infection. For example, the recruitment of histone deacetylases (HDACs) at BZLF1 and EBNA2 Cp promoters maintain EBV latency (46) (47, 48) and, as expected, HDAC inhibitors, such as sodium butyrate, can activate the EBV lytic cycle (49, 50). Of note, EBV DNA within virions or soon after lytic replication is nucleosome free, potentially to allow its encapsulation into the nucleocapsid (44, 51). DNA methylation is typically associated with transcriptional silencing (52, 53). The EBV genome is known to be hypermethylated in the latent form and in virions (54). DNA methylation plays a critical role in transcriptional regulation of LMP1 and EBNA2 and thus contributes to the transition among latency programs (55). Consistently, inhibitors of DNA methylation, such as 5-azacytidine can reactivate latently infected lymphoblastoid cells (LCLs) (56). However, since the OriP region is required for EBV transcriptional regulation, it is typically depleted of DNA methylation.
EBV typically exhibits dual tropism with the capacity to actively infect and replicate both in epithelial and B-cells. Sometimes EBV can also infect other targets such as T lymphocytes and natural killer (NK) cells (7, 57). The entry of EBV into target cells is facilitated by its envelope glycoproteins (gp). B-cell entry requires glycoproteins gp350, gH, gL, gB and gp42, whereas epithelial cell entry needs BMFR2, gH, gL and gB (58–61). In epithelial cells, EBV is more likely to be transferred from EBV-positive B lymphocytes that cause lytic infection (62). T-cell and NK-cell entry seem to also require gp350 and gp42, respectively (63, 64). The EBV virion has a diameter of 150–170 nm, consists of a linear, ~172 kbp double stranded DNA that codes for more than 85 protein coding genes (65, 66). However, the exact function of 30-40% of these genes remains unknown (67). The EBV genome also has several tandem repeat regions that serve various functions (20, 68). The entire genome is enclosed within an icosahedral capsid surrounded by a layer of tegument proteins and a lipid envelope that is made up of several unique glycoproteins. EBV can enter human B cells via a high-affinity interaction between viral gp350 and host complement receptor type 2 (CR2) protein. HLA class II can act as a co-receptor (15, 69). These protein-protein interactions stimulate endocytosis of the virus into non-clathrin coated vesicles of B cells. The virus also infects epithelial cells as well as T- or NK-lineage cells albeit at a lower frequency. Ephrin Receptor A2 (EphA2) was recently identified as the entry receptor for EBV in epithelial cells. This protein interacts with EBV glycoproteins gH/gL and gB (70). Although less is known about the mechanisms of EBV entry into other cells, CR2 has been identified as the entry receptor for T lymphocytes (64) but is apparently not essential for entry into NK cells (71). HLA class II also plays a role for entry into NK cells but its role for T cells remains less clear (63). Upon entering B cells, the viral genome typically persists in the nucleus as a circular episome, expressing a subset of genes that promote survival of the infected host cell (58, 59). Typically, after initial infection, the EBV genome rapidly circularizes either before or at the same time as the initial phase of viral mRNA synthesis (21, 72–74). After B cell infection, EBV initiates an often asymptomatic, lifelong latency program in a few cells with extremely low viral activity. During this stage, the EBV episome is replicated by the host cell DNA polymerase primed on the EBV origin of plasmid replication (OriP) (75, 76).
EBV was originally divided into two major sub-variants, type 1 and type 2, based on the sequence of two EBV-encoded genes - EBNA2 and EBNA3 (77, 78). Type 1 is prevalent globally (e.g., B95-8, GD1, and Akata strains) while type 2 (e.g., AG876 and P3HR-1 strains) is endemic to sub-Saharan Africa (79). Currently, more that 71 distinct EBV strains have been identified. EBV variants have different replicative properties and individuals may become infected with two or more strains. With the advent of high-throughput sequencing technologies, it is now possible to sequence EBV genomes from clinical specimens of diverse populations with different malignancies. The first sequenced genome was of the prototypical EBV B95-8 strain. This strain harbored a 12-kb deletion in its genome. It was not until 2014 that this defect was noted and EBV from Raji strains was recovered to get the final complete sequence of wild-type EBV (EBV-wt,26 GenBank accession no. NC_007605.1) (65). This is now the gold standard reference sequence for many research groups in the field. It has been reported that certain EBV strains have more oncogenic potential than others. For instance, the NPC derived EBV strain, M81, spontaneously replicates at an unusually high rate in B cells and has an extremely high propensity to infect epithelial cells (80). This ‘super infectious’ property is attributed to a single nucleotide polymorphism (SNP) in the BZLF1 promoter region that confers binding by host cellular transcription factors, notably NFATc1 (81). Increasing studies are now investigating the heterogeneity of EBV latent and lytic genes among the different EBV strains in order to identify high-risk EBV strains (79). Doing so will potentially help identify high-risk infected individuals and facilitate development of effective EBV vaccines and anti-EBV T-cell therapies.
EBV latent proteins are generally considered key drivers of tumorigenesis in EBV-associated cancers and thus it is important to understand their functions in establishment of persistent infection and cellular transformation. In this section we will briefly describe the function of EBV-encoded latent gene products and their role in transformation and oncogenesis in EBV-associated malignancies.
EBNA1 is a transcription factor that is essential for EBV episomal maintenance and replication (82, 83). Consistently, EBV variants that harbor EBNA1 deletion do not have the capacity to establish episomal latent infection (84). The DNA binding domain of EBNA1 is necessary but not sufficient for EBV replication and requires the N-terminal region (85). Since EBNA1 lacks enzymatic activity, it primarily recruits host cellular factors to replicate EBV episomes and to govern mitotic segregation (86). Of note, EBNA1 can also function as a transcriptional repressor and can downregulate its own transcription in an autoregulatory loop (87–89). In terms of oncogenic potential, EBNA1 is involved in progression of carcinogenesis. Specifically, EBNA1 deletion significantly decreases immortalization efficiency, while its overexpression inhibits apoptosis (90, 91). EBNA1 modulates several cellular signaling pathways that provide survival advantage to infected cells (92). While EBNA1 is reported to enhance phosphorylation of STAT1 in one gastric cancer cell line and two nasopharyngeal cancer cell lines, it inhibits anti-tumor TGF-β1 and NF-κB pathways, promoting tumorigenesis (93, 94). EBNA1 also upregulates several proteins involved in metastasis and oxidative stress in EBV+ NPC cells (95). In addition, EBNA1 induces loss of promyelocytic leukemia (PML) nuclear bodies and subsequently abrogates PML functions, such as p53 activation and apoptosis, resulting in increased survival of gastric cancer cells (96). The fact that EBNA1 is the only EBV protein that is consistently expressed in all latency types, and therefore in all EBV-associated tumors, makes it a key target for EBV specific therapies. Consistently, pharmacological inhibition of EBNA1 using a small-molecule inhibitor VK-1727 has been tested in various in vivo xenograft mouse models for specific EBV+ cancers. These studies have demonstrated that inhibition of EBNA1 can selectively suppress EBV+ tumor cell proliferation (97).
EBNA2 is a transcriptional activator of both cellular (e.g., CD21, CD23 and c-MYC) and viral genes (e.g., LMP1 and LMP2) (98, 99). EBNA2 plays a crucial role in transcriptional reprogramming of B cells to facilitate growth and survival (100, 101). However, unlike EBNA1, it cannot bind to DNA directly and requires host transcription factors, such as the Notch pathway DNA-binding factor RBP-Jκ and PU.1 to regulate gene transcription (102). In terms of oncogenic potential, EBNA2 plays a crucial role in the transformation process and functionally mimics Notch (103, 104). Consistently, P3HR-1, a variant EBV strain in which EBNA2 and the last two exons of EBNA-LP are deleted, does not transform B cells in vitro. EBNA-LP, a transcriptional co-activator of EBNA2, is also an important EBV oncoprotein that drives B cell transformation and functions by up-regulating the expression of EBNA2 targets (105). In addition, EBNA2 activates MYC enhancers via long-range interactions. MYC can both increase proliferation and sensitize cells to apoptosis. However, it is also a known proto-oncogene, so unsurprisingly, EBNA2-mediated MYC activation seems to promote lymphomagenesis in Burkitt lymphoma (106).
The EBNA3 protein family members are stable, tightly regulated and consist of EBNA3A, EBNA3B and EBNA3C (107). EBNA3 proteins are well studied classes of transcriptional regulators known to regulate both EBV (e.g., LMP1) and host gene (e.g., CD21) expression and, depending on context, can function as activators or repressors of gene expression (108–110). All EBNA3 proteins play a role in transformation and prolonged persistence of EBV in infected B cells. EBNA3 transcripts are generated from the Cp latency promoter and are reported to be only expressed in B cells as a part of the latency III program (111). Like EBNA2, EBNA3 proteins also do not directly bind to DNA but are, rather, recruited by cellular DNA binding factors, such as RBP-Jκ (112, 113). RBP-Jκ tethers EBNA3s to chromatin, but binding of EBNA2 and EBNA3 to RBP-Jκ are mutually exclusive (113, 114). In terms of their oncogenic potential, the EBNA3 family of proteins have antagonistic functions but cooperate in a complex to facilitate EBV persistence, as well as to promote oncogenic transformation. EBNA3A and EBNA3C are considered oncogenes since they are also involved in B cell transformation. However, in the absence of EBNA3A and EBNA3C, EBV can latently persist in humanized mice (115). Despite the sequence and structural similarity and reports about their functional cooperativity, EBNA3 proteins are often have opposing functions. For example, EBNA3B is dispensable for B cell transformation but inhibits EBNA3A- and EBNA3C-mediated oncogenic functions in vivo. In addition, like EBNA2, EBNA3 has long-range interactions with enhancers and super-enhancer elements that drive oncogenesis (116).
EBV encoded LMP1 is an essential membrane protein that has an N-terminal cytoplasmic tail, six transmembrane domains and a C-terminal cytoplasmic region that is divided into two C-terminal activation regions 1 and 2 (CTAR1 and CTAR2). These regions are required for tethering LMP1 to the plasma membrane and its signaling activity (117).. LMP1 mimics cellular CD40 receptor, a member of the (TNFR) superfamily and can drive growth and differentiation of B cells by substituting CD40 functions in vivo (118). LMP1 signaling is primarily mediated by the ability of host TNFR-associated factors (TRAFs) or death domain protein TRADD to interact with CTAR1 or CTAR2 to facilitate activation of upstream regulators of several signaling pathways (119, 120). In terms of oncogenic potential, LMP1 is a well-documented EBV oncogene and is essential for transformation of B cells in vitro. LMP1 acts as a constitutively active CD40 receptor and thus can activate target signaling pathways (e.g. the NF-kB pathway) independent of ligand engagement (121–123). These includes pro-tumorigenic functions, such as increase in cell proliferation, cytokine production (IL-6, IL-8), apoptotic resistance (by upregulating the expression of anti-apoptotic protein (Bcl-2, A20)), immune modulation, anchorage-independent growth, metabolism, angiogenesis, metastasis and invasion, all of which are known to contribute to EBV-mediated pathogenesis (124–126).
The LMP2 gene encodes two dominant isoforms, LMP2A and LMP2B. LMP2B is the smaller isoform that structurally lacks a short cytoplasmic N-terminal domain that harbors an essential survival signal known as immunoreceptor tyrosine-based activation motif (ITAM) (75, 127). Interestingly, B cell receptor (BCR) also has an ITAM motif which, upon phosphorylation, recruits and activates the Src family and Syk protein tyrosine kinases (PTKs) and promotes B cell proliferation and differentiation. However, the association of these PTKs with the phosphorylated ITAM of LMP2A negatively regulates PTK activity (127), thereby inhibiting BCR-driven calcium flux, tyrosine phosphorylation and BZLF1 induction in LCLs (128). In terms of oncogenic potential, LMP2A and LMP2B seems to be dispensable for in vitro B-cell transformation (129). Indeed, LMP2 has anti-oncogenic potential (130, 131).
In addition to the latent proteins mentioned above, there are a few other EBV-encoded latent gene products including EBERs, BARTs and EBV microRNAs. Two EBERs (EBER 1 and 2) are small non-coding RNAs that are expressed during all latency programs. In terms of oncogenic potential, EBERs do not seem to be essential because EBER-deleted EBV strain can similarly transform primary B-lymphocytes (132). However, they can affect cellular processes by enhancing anti-inflammatory cytokine IL-10 production via RIG-I/IRF3 activation (133). In certain B cell lymphomas with restricted type 1 latency, EBER2 and EBNA1 can induce expression of cytokines (e.g., IL6) or cytokine receptors (e.g., CD25) to promote B cell survival (134). The BARTs encode BARF0, RK-BARF0, A73 and RPMS1. The function of BART proteins encoded by corresponding ORFs needs further examinations. The 44 EBV microRNAs are arranged either adjacent to the BHRF1 gene or within the BART introns. These microRNAs are associated with different EBV latency programs (135, 136) and are differently induced during the lytic cycle (137). However, the significance and function of most remain unclear.
EBV is associated with a wide variety of diseases and malignancies. Infectious mononucleosis (IM) is an extremely common, self-limiting, and acute disease associated with primary EBV infection. It is characterized by lymphadenopathy, transient fever and hepatosplenomegaly that usually resolves in time. Chronic active EBV infection (CAEBV), although rare, is a severe and fatal condition characterized by unusually high EBV DNA load (103–107 copies/mL) (138), which is now considered to be one of the EBV+ T or NK cell lymphoproliferative diseases and can lead to two lethal conditions: hemophagocytic lymphohistiocytosis and chemotherapy-resistant lymphoma (139). Historically CAEBV was partially managed using immunomodulatory agents such as interferon-α (IFN-α) and IL-2 (140), but JAK/STAT inhibitors are now a standard component of treatment (141). EBV is also a major risk factor for immunocompromised patients. In HIV patients the lack of efficient and EBV-specific T cell responses significantly increases the risk of developing EBV-associated lymphoma (142, 143). Oral hairy leukoplakia (OHL) is a hyperproliferative disorder observed in immunocompromised patients that is triggered by EBV lytic state (144–147). Post-transplant lymphoproliferative disorder (PTLD) represents severe, life-threatening uncontrolled B cell proliferation post-organ or bone-marrow transplantation, which in majority cases is associated with EBV reactivation and replication (148). EBV infection is typically associated with cases of early onset (within one year of transplantation) compared to late-onset PTLDs (149) and higher EBV viral loads increase the risk of PTLD (150). One of the main reasons for EBV reactivation during transplantation is the use of immunosuppressants to prevent transplant rejection. However, these immunosuppressants inhibit all T cells, including EBV-specific ones, providing an opportunistic means for EBV to escape from immune surveillance (151). Consistently, pre-emptive treatment with inhibitors of EBV DNA replication can reduce the incidence of PTLD (152). EBV infection has also been implicated in the development of autoimmune diseases, such as multiple sclerosis (MS) (153). MS is characterized by autoreactive B cells in the cerebrospinal fluid (CSF) that attack the myelin sheath of the central nervous system (CNS). Recently, a large-cohort study on 62 million serum samples taken from over 10 million US military personnel provided compelling evidence suggesting a necessary but not sufficient role of EBV infection towards the development and progression of MS (154). Pathologically, another recent study showed that EBNA1 mimics the CNS protein glial cell adhesion molecule (GlialCAM), which is expressed by myelin sheath-forming cells. Antibodies against a particular region of EBNA1 highly cross-react with GlialCAM in MS patients, potentially resulting in “off-target” autoimmune attack against the myelin sheath in CNS of patients with MS (155). The ability of EBV to immortalize B cells is testament to its tumorigenic potential (156, 157). Indeed ~1-2% of all human tumors are attributed to EBV, equating to ~300,000 new cases worldwide in 2020 (158–160). EBV infects both genders, however, EBV-associated malignancies are slightly more prevalent in males compared to females (161). EBV infection is associated with various lymphomas, including Burkitt’s lymphoma (BL), Hodgkin lymphoma (HL), diffuse large B cell lymphoma (DLBCL), NK/T cell lymphoma and primary effusion lymphoma (6, 162), as well as epithelial malignancies, such as NPC and gastric carcinoma (GC). Below, we will discuss some EBV-associated malignancies in greater detail.
BL is a highly aggressive B cell non-Hodgkin neoplasm first reported in Africa by Denis Burkitt (163). The WHO classification describes three clinical variants of BL: endemic (eBL), sporadic (sBL), and immunodeficiency-related (usually HIV-related) (164). While around 95% of eBL are EBV+, only 15% of sBL and 40% of immunodeficiency-related BLs are associated with EBV (67). Despite primarily having the type I latency programs, some other EBV genes (e.g., EBNA2) are sporadically detected (165). Although the role of most EBV genes in BL pathogenesis remains unclear, it has been shown that EBNA1 can inhibit apoptosis in BL cell lines by interacting with host proteins, such as p53-regulator USP7 and an anti-apoptotic protein survivin (166). The eBL variant is common in malaria endemic regions and is commonly characterized by the presence of large tumors in the head and abdominal cavity. EBV is detected in nearly all the cells of eBL tumors, however the exact underlying mechanism has not yet been fully elucidated (167). Fortunately, BL tumors typically respond to intensive chemotherapy and are often curable when diagnosed early but it still remains a fatal disease in much of the affected sub-Saharan African population (168). This is attributed to several factors, such as diagnostic delay, inadequate healthcare, poverty, and malnutrition (169, 170). Outside of malaria-endemic regions, the occurrence of BL is about 10-fold lower mostly constitutes the sBL variant, which is concurrent with a lower EBV prevalence (10-30%) (171). Clinically, the majority of sBL cases present as tumors in the abdominal and thoracic cavities. Despite a poor prognosis, current ongoing clinical trials using a modified chemotherapeutic approach are showing some promise (NCI 9177 trial). Nonetheless, more specific and less toxic treatment options are needed. The best known molecular feature of BL is the translocation of the proto-oncogene MYC to an enhancer locus next to the immunoglobulin heavy chain gene, causing constitutive expression of MYC (172, 173). However, in addition to enhanced MYC activity, the development of BL requires additional genetic or epigenetic aberrations (174). Over the years, extensive genomic and transcriptomic characterization of BL cases have identified genes that are recurrently mutated (e.g., BCR, TCF3, ID3, CCND3, ARID1A and SMARCA4) (175, 176). For instance, it has been experimentally demonstrated that mutations in ID3 promote proliferation and cell cycle progression (177). Genes in the ID3-TCF3-CCND3 pathway are frequently mutated in MYC-rearranged eBLs and may represent one of the major underlying causes of BL (178). Nevertheless, the mutational and transcriptional landscape of EBV+ BLs is quite distinct from EBV– BLs and is primarily attributed to the presence of EBV. Recent studies have explored the spectrum of aberrations in EBV+ BLs and the complex interplay between specific viral-host transcriptional programs (179, 180). For example, the frequency of MYC, ID3, TCF3 and p53 somatic mutations is lower in sBLs, while the frequency of mutation in ARID1A, RHOA and CCNF are higher in eBLs (179). Interestingly, it has been previously reported that LMP2A enhances MYC driven lymphomagenesis through activation of the PI3K-pathway (181–183), suggesting that activation of PI3K by LMP2A might be an alternative and/or convergent mechanism to the one driven by TCF3/ID3 mutations. In addition to LMP2A, a recent study described the role of LMP1 in MYC-induced lymphomagenesis in a subset of BL cases (184). Further, comparative transcriptome analysis of eBL and sBL tumors have highlighted key mutational differences between the two types of BLs, with sBL having a significantly higher mutational burden, which correlates strongly with the EBV status rather than geographical distribution (185, 186). A recent study has found a genome-wide increase in aberrant somatic hypermutation (SHM) in EBV+ BLs, attributed to the higher expression of host activation-induced cytidine deaminase gene AICDA, which is also a target of EBNA3C (187). Overexpression of AICDA increases the likelihood of DNA breaks and MYC translocations as well as pathogenic mutations (188). Other factors, such as co-infection, seem to also contribute to BL pathogenesis. For example, Plasmodium falciparum induces DNA damage, which can turn EBV-infected B cells into eBL. Likewise, impaired immune surveillance in HIV-infected patients can induce EBV-associated BLs (179, 189, 190).
HL is a lymphoid neoplasm that originates from B cell. The two major forms of HL are the classical type (cHL) and the nodular lymphocyte predominant type (NLPHL), the latter being considered as an EBV– malignancy. One of the main features of cHL is the presence of large multinucleated cells known as Hodgkin and Reed-Sternberg (HRS) cells (191). Although derived from B cells, HRS cells lack the normal B cell phenotype which is attributed to functional aberrations in key B-cell associated TFs, such as PAX5, EBF1, TCF3/E2A and NF-κB (192, 193). In addition, the HRS cells co-express various hematopoietic cell markers and have anomalous activation of several signaling pathways (e.g., NF-κB and JAK/STAT), attributed to the frequent mutations of key TFs and/or cellular interactions within the tumor microenvironment (TME) (194). Globally, nearly 50% of cHL cases are EBV+ but the EBV prevalence varies with geography. For instance, about 30-40% of cHL cases in North America and Europe are EBV+, while in Africa, Asia, and Latin America, all cases are EBV+ (192). EBV+ cHLs are typically characterized by a massive immune cell infiltration (195). Although the exact mechanistic role of EBV in cHL pathogenesis is unclear, the presence of EBV throughout disease progression underscores its role in maintaining the tumor phenotype (196). EBV+ cHLs exhibit type II latency program, maintaining high levels of LMP1 and LMP2A proteins in all HRS cells (191, 197). LMP1 and LMP2A can both contribute to the pathogenesis of cHLs by mimicking cellular receptors, namely CD40R and BCR, that are essential for cell survival and expansion (99, 198–204).
NPC is a unique and complex form of a head and neck epithelial cancer. While the disease prevalence is extremely low in the Western world, it is endemic in Asia, Southeast Asia, North Africa, Greenland, Alaska, and the Middle East, affecting around 30 per 100,000 individuals. The distinct geographical distribution pattern of NPC cases worldwide suggests both environmental (e.g., consumption of preserved, salt-cured foods) and genetic factors (e.g., mutations in HLA, TNFRSF19, CDKN2A/B, and TERT) as its etiology (205–208). Nonetheless, EBV infection is reported to be a critical risk factor and plays an essential role in NPC progression (209–211). About 90% of malignant cells in NPC are either undifferentiated or poorly differentiated squamous epithelial cells that typically express several EBV latency type II gene products (212). These include EBER1/2, EBNA1, LMP1, LMP2, BARF1, and several other EBV-encoded non-coding transcripts. LMP1 is one of the key oncogenic drivers of NPC that is expressed in 20%–60% of NPCs and all pre-malignant or pre-invasive lesions, making it a prime therapeutic target (213). NPCs harbor a high somatic mutation burden. A recent study identified more than 50 mutations per tumor in a panel of 111 micro-dissected EBV+ tumor samples. A whole exome sequencing study of NPCs identified a range of somatic mutations in key cellular genes and pathways including p53, HLA, NF-κB, MAPK, and P13K (214). Given that somatic mutations in NF-κB pathway were mutually exclusive to LMP1-overexpressing NPCs, the NF-κB pathway activation either by EBV or mutation seems to be vital for NPC pathogenesis (215). This is corroborated by another genome-wide analysis that reported that ~90% of the EBV+ NPCs have constitutive activation of NF-κB inflammatory pathways either due to somatic mutations or expression of EBV encoded LMP1 oncogene (216). Additionally, chromosome instability (CIN) is hallmark of NPCs. Early studies have linked EBV infection with genomic instability. Detailed differences in the genomic and epigenomic landscapes of EBV driven epithelial malignancies have been reviewed elsewhere (207).
Gastric cancer (GC) is one of the leading causes of cancer-related mortality (165). Nearly 10% of the 950,000 yearly new cases of GC cases are attributed to EBV infection. EBV+ GC usually mimics the histological features of lymphoepithelioma-like carcinoma, in which dense lymphocytic infiltrates (mainly CD8+ T cells) are present. EBV+ GC is, in fact, of the four molecular subtypes of GC, namely EBV+ GC, GC with microsatellite instability, genomically stable GC, and GC with chromosomal instability (217). EBV+ GC and GC with microsatellite instability are mutually exclusive. There are two common theories regarding the mechanism of occurrence of EBV+ GC. First, that EBV enters the digestive tract via the saliva and directly infects gastric epithelial cells. Second, EBV within B cells of the stomach is reactivated (through unknown mechanisms) and infects surrounding gastric epithelial cells (218). EBV+ GC cells exhibit latency I or intermediate latency I/II programs (219, 220). Consistently, EBNA1 and LMP2A are expressed in 100% and ~50% of EBV+ GC cases, respectively, but LMP1 is not expressed (221). EBV+ GCs have distinct genomic aberration, clinicopathological features, cellular gene methylation, and comparatively favorable prognosis compared to EBV– GCs (217, 222–224). Unlike NPCs and EBV– GCs, p53 mutations are rare in EBV+ GCs (225). This might also partially explain a comparatively favorable prognosis for EBV+ patients since it is known that mutations in p53 reduces sensitivity to chemotherapy and radiation (226). EBV can also extensively induce cellular DNA methylation, which could inhibit tumor suppressor genes (e.g. p16 and E cadherin) and thus increase the risk of cancer formation (227). Recent studies report the increased expression of certain immune checkpoint proteins in EBV+ GC, such as PD-L1 and IDO-1 and their upstream regulators (180), which could explain their favorable response to immune checkpoint therapy (228). The pathogenic role of EBV, underlying molecular mechanisms and current treatment options for EBV+ GC have been further discussed elsewhere (218).
To establish infection and persistence, EBV employs different strategies to evade the host immune response and to compromise innate and adaptive immunity during its life cycle (229). Some of these mechanisms includes genetic and epigenetic alterations, inhibition of apoptosis, enhanced cell proliferation and inhibition of immune recognition of EBV-infected cells ( Figure 2 ). Broadly, tumorigenesis can occur by i) enhancing antiapoptotic or reducing proapoptotic gene expression; ii) promoting cell growth and survival signaling pathways; and iii) shaping the tumor microenvironment for malignant cells to escape immune surveillance. In this section, we will briefly discuss some of these mechanisms, focusing on the role of the tumor microenvironment and immune escape mechanisms in EBV-induced malignancies.
EBV is known to alter the cellularity and the properties of the tumor microenvironment (TME) thereby shaping an immunosuppressive environment. This involves inhibition of anti-tumor effector immune cells, such as NK cells and CD8+ T cells and recruitment and differentiation of immune-suppressive and/or anti-inflammatory cells, such as regulatory T cells (Tregs), dendritic cells (DCs), Th17 cells, M2-polarized tumor-associated macrophages (TAMs) and myeloid-derived suppressor cells (MDSCs) (230). In addition to the immune compartment, TME also includes stromal cells, soluble mediators such as chemokines and cytokines that can be modified by EBV. Together, these changes facilitate tumor growth by several mechanisms including promoting immune evasion. Extensive research has been done to understand the TME of both EBV– and EBV+ malignancies (67). Studies on the TME are limited for certain EBV+ malignancies due to low incidence rates (e.g. NK/T cell lymphoma) or high heterogeneity among the sites of the disease and immune compartment (e.g. PTLD). It is known that tumor cells can affect immune cell infiltration as well as drive the infiltrated immune cells towards a tolerogenic or exhausted state rendering them non-functional. Although highly variable, EBV+ carcinomas are generally characterized by high immune cell infiltration. This includes CD8+ T cells, CD4+ T cells (Th1, Th2, Treg cells, etc.) and CD163+ M2-TAMs (TAMs) (231). T cells are prevalent in EBV+ epithelial cancers. For example, EBV+ GCs attract high numbers of CD8+ T cells better known as cytotoxic T lymphocytes (CTLs) (232) and the CTL infiltration is positively correlated with EBV viral load (233). Despite a significant increase of CD8+ T cells within the TME of EBV+ NPCs, they exhibit an exhaustion signature and reduced cytotoxic activity (234, 235). In EBV+ NPC, LMP1-mediated glycolysis promotes MDSC expansion within TME leading to tumor-induced immunosuppression (236). Tregs and the CD8+ T cells are dominant in cHL. However, CD8+ T cells are primarily exhausted due to the high levels of PD-L1 expression (237). In contrast, higher numbers of M2-TAMs seems to be the most prominent in the TME of Burkitt lymphomas (238), which are also known to affect tumor progression via upregulation of immune checkpoints and expression of specific cytokines. Similarly, an increased frequency of CD57+ NK cells is reported in EBV+ GC, NPCs and cHL compared to their EBV– counterparts (231). Unlike T cells, the role and the presence of B cells within the TME of EBV+ malignancies are mixed and warrant further investigation. Altered expression profile of certain soluble mediators including cytokines within the TME have an important role in EBV-associated malignancies and often these alterations precede immune cell infiltration. For instance, EBV induces host CXCL9, CXCL10, and CCL20 in some EBV+ tumors, which in turn attract regulatory T cells into TME (239, 240). The cytokines often have pleiotropic effects. For instance, IL-10 is known to downregulate the expression of HLA class I and II antigens, induce Tregs (which in turn inhibit T-cell proliferation and IFN-γ secretion) and inhibit CD8+ T cell cytotoxic function resulting in an overall immune-suppressive microenvironment within the tumors (241–243). Other soluble mediators such as IL-1β, IL-4, IL-6, IL-8, and IL-13, IFN-γ, CXCL10 and CXCL12 are also frequently upregulated in EBV+ malignancies and implicated in disease progression (231). It is also known that EBV proteins such as LMP1 and EBNA1 can significantly promote an immunosuppressive microenvironment by promoting expression of chemokines and cytokines. Compared to their EBV– counterparts, EBV+ GC cells have an overall higher involvement of Th1 and CD8+ T cells and produce more cytokines/chemokines including CCL20, CCL22, and CCL17 (244, 245). Other non-immune cells within the TME (e.g., cancer-associated fibroblasts) are also known to produce pro-inflammatory cytokines and have been reported to surround tumor cells in EBV+ solid tumors (246).
The human immune system has developed several strategies to combat invading pathogens. Central components of such strategies are the innate and adaptive immune responses. While innate immune responses are the first line of defense, they are often non-specific. In contrast, the adaptive immunity is more specific and long lasting and maintains specific memory of invading pathogens. Despite these host immune defense tools, EBV can establish latency within infected cells suggesting that the virus has developed mechanisms to escape, inhibit, or subvert host immune responses to ensure its own persistence. A typical mechanism of innate immune evasion in EBV infection is downregulation of pattern recognition receptors, such as toll like receptors (TLRs). Similarly a general mechanism of adaptive immune-evasion in EBV-associated malignancies is the overexpression of immune checkpoint proteins (e.g., PD-L1, IDO-1, CTLA-4, LAG-3, TIM-3, and VISTA), thus making them susceptible to treatment with immune checkpoint blocking immunotherapy (247). Below, we will briefly discuss the strategies used by EBV to evade the innate and adaptive immune responses.
The innate immune response against EBV originates from both the EBV-infected cells themselves (B and epithelial cells) as well as from bystander cells like myeloid and NK cells. One of the major elements of innate immunity are the pattern recognition receptors (PRRs) that can recognize a diverse array of pathogen associated molecular patterns (PAMPs) and recruit downstream effector mechanisms, such as secretion of type I interferons in response. To date, 10 TLRs have been identified in humans. TLR9 is the key receptor for sensing EBV and is abundantly expressed in B cells and certain myeloid cells. TLR9 specifically senses unmethylated CpGs of EBV DNA motifs present in viral particles immediately after primary infection in B cells. Upon stimulation, TLR9 can activate the NF-κB pathway, which in turn promotes production of pro-inflammatory cytokines and B cell proliferation. Dendritic cells (DCs) can sense, phagocytose, process and present antigens to cells of the adaptive immune system. DCs are generally classified into two types, conventional DCs (cDCs) and plasmacytoid DCs (pDCs), which express TLR3 and TLR9, respectively. Unlike TLR9, TLR3 is endosomally located and recognizes dsRNAs, including EBV-RNAs (EBERs). Nevertheless, both TLR3 and TLR9 stimulate type I IFN production when triggered. Even monocytes and macrophages sense EBV via TLR3 and TLR9 leading to cytokine and chemokine production. Another interesting cellular player are the NK cells. NK cells are critical cytotoxic innate lymphocytes that target infected cells. Like DCs, NK cells have two broad subsets, CD56bright and CD56dim, the latter being more relevant in B cells restricted EBV infections (248). This is supported by studies which suggest that deficiencies in NK cells can increase the occurrence of EBV-driven pathologies. Consistent with this notion, NK cells recognize and preferentially target infected cells with lytic EBV infection (249, 250). Despite the intricate network of innate immune players, EBV has developed strategies to counteract innate immunity. For example, EBV can reduce expression of several TLRs. For example, EBV lytic protein BGLF5 reduces TLR9 expression and LMP1 suppresses TLR9 function in EBV+ PTLDs and cHLs. A detailed review of interplay between EBV and host innate immune responses can be found elsewhere (251).
Adaptive immunity can be broadly classified into humoral and cell-mediated processes, which are mediated primarily by B and T cells, respectively. Adaptive humoral responses are the direct product of interaction between antigens and immunoglobulin (Ig) on the surface of naïve B cells, which leads to secretion of antigen-specific antibodies and antigen presentation to T cells. T cells in turn help B cells with Ig-class switching and affinity maturation either specifically via CD40L/CD40 binding or non-specifically via interleukin/cytokine release (252). Primary EBV infection triggers an immediate IgM response to viral capsid antigen (VCA) and BMRF1 encoded early antigen diffuse (EaD) complexes. The importance of humoral immune responses and molecular details of antibody response against EBV has been previously reviewed (252). EBV-specific T cells are key players in determining the fate of EBV infected cells. Both types of T cells, namely, CD8+ cytotoxic and CD4+ helper T cells can recognize EBV antigens presented on the surface of infected cells by HLA molecules (253, 254). While HLA class I antigens are expressed on almost all nucleated cells, HLA class II antigens are expressed on the surface of antigen presenting cells (APCs) (253). Despite increased infiltration of CD8+ T cells in EBV+ tumors compared to EBV– tumors (255), EBV has developed several strategies to evade T cell responses, for example by downregulating HLA expression, blocking antigen presentation pathways or creating an immunosuppressive TME. The latter is mediated by increased production of immunosuppressive cytokines and/or increased expression of immune checkpoint molecules that are known to induce T cell exhaustion (229). Glycoprotein programmed death ligand 1 (PD-L1) represents one of the several immune checkpoint molecules that is modulated by EBV and is used as a mechanism of immune evasion by many tumors. This occurs as a function of PD-L1 engagement with cell surface receptor programmed death 1 (PD-1 or CD279) expressed on T cells (256). PD-1 interacts typically with its ligand PD-L1 (CD274 or B7-H1) and less frequently with PD-L2 (CD273 or B7-DC) (257). While PD-L1 is expressed by a wide-range of cell types, the latter is expressed only by specific cell types, including DCs, mast cells and macrophages (258). PD-1 is an inhibitory receptor that is rapidly upregulated upon antigen-mediated T cell receptor (TCR) stimulation (259). Recent studies have identified PD-1 expression in a plethora of other immune cell subsets such as B cells, DCs, NK cells, and monocytes (260, 261). The PD-1/PD-L1 pathway plays a crucial role in immune tolerance by fine-tuning the quality and duration of T cell response thereby serving as a ‘rheostat’ of immune response (262, 263). This is partly achieved by counterbalancing the T cell activation signal triggered by binding of CD28 on T cells with CD80/CD86 on APCs. The binding of PD-1 receptor with its cognate ligands attenuates TCR signaling and leads to T cell exhaustion (264). This inhibitory interaction serves to protect target tissues from hyper-activated immune mediated damage. Tumor cells take advantage of this mechanism by frequently upregulating PD-L1 expression to escape the host anti-tumor immune response (265). Consequently, the PD-1/PD-L1 axis serves as one of the promising targets for immunotherapy in such malignancies. Infectious viruses such as Epstein-Barr virus (EBV), hepatitis C virus (HCV) and hepatitis B virus (HBV) leverage the PD1/PD-L1 pathway to facilitate escape of infected cells from the antiviral immune response (266). Consistently, PD-L1 expression is higher in EBV+ relative to EBV– tumors in NPC, GC and DLBCL (180). Several mechanisms have been reported for increased PD-L1 expression. These include alterations at the genetic level, specifically the amplification of the chromosomal region 9p24.1, which includes the genes PD-L1, PD-L2, and JAK2 (217, 267). Such genetic alterations have been associated with certain B-cell lymphomas and gastric cancer (268–271). Interestingly, in ~40% of cHLs, increased PD-L1 expression is not due to this amplification but attributed to upregulation by certain EBV-encoded gene products (272). The dysregulated expression of PD-L1 in cancer has been attributed to the oncogenic activation of multiple signaling pathways, including JAK/STAT, PI3K/Akt/mTOR, MEK/ERK, and Jun/AP-1 which can either act independently or synergistically to regulate PD-L1 expression (273–275). In addition, another important mechanism of PD-L1 upregulation is 3’-UTR disruption of PD-L1 by EBV insertion at this locus (276). EBV encoded genes are also known to modulate host immune responses. BZLF1 (Zta) can induce expression of host immunosuppressive genes, such as TGFB1, which further downregulate expression of immune responsive genes, such as TLR9, IFI6, and IL23A (277, 278). LMP1 promotes AP-1, JAK-STAT and NF-κB signaling mediated activation of PD-L1 (279–281), suggesting that LMP1-mediated signaling might also be a key player in the immune escape strategy in cancers that express LMP1 (e.g. NPC, cHLs and DLBCLs). LMP1 also promotes proliferation and survival and LMP1-driven PD-L1 upregulation correlates with poor prognosis in certain lymphomas (282). Lack of LMP1 expression in EBV+ eBLs, consequently, is associated with absence of PD-L1 expression observed in these tumors (283). Likewise, EBNA1 also modestly promotes IFN-γ-induced PD-L1 overexpression in GC cell lines (284). The role of other EBV genes in inducing PD-L1 expression in EBV-associated cancers is less clear and further investigation is needed to determine how different viral gene products affect immune responses and PD-L1 expression in different EBV-associated cancers.
Since EBV contributes to malignant cell transformation and is found in almost every cell of EBV+ tumors, it has been considered a potential target for precision medicine and individualized cancer treatment. While non-specific chemotherapy is typically the first line of therapy, several additional strategies have been proposed specifically targeted towards EBV+ malignancies ( Figure 3 ). These include i) antivirals against EBV; ii) small molecule inhibitors for EBV-encoded gene products, such as LMP1 and EBNAs; iii) induction of the lytic form of EBV replication in tumor cells in combination with prodrugs that are cytotoxic in lytically infected cancer cells; iv) enhancing the host immune response to viral antigens expressed by EBV-infected tumor cells; v) use of EBV vaccines. Additional strategies that are under consideration include induction of EBV episome loss by treating tumors cells with low-dose hydroxyurea and expressing toxic genes using EBV-dependent approaches (285, 286). There is also considerable enthusiasm for immune checkpoint therapies for the management of EBV associated cancers. Some are approved or under investigation in clinical trials for the treatment of NPC, GC, and HL. Based on initial trial reports, PD-1 targeting treatments, such as pembrolizumab and nivolumab, seem to improve longevity and/or partial response, especially in patients with PD-L1+ tumors (287–293). Here, we will discuss some of the major therapeutic strategies for EBV+ cancers, focusing on recent developments and highlighting current gaps and/or challenges.
For lymphomas, such as BLs, the conventional chemotherapeutic regimen include R-CHOP (Rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone), CALGB (cancer and leukemia group B), Hyper-CVAD +/- R (cyclophosphamide, vincristine, doxorubicin, and dexamethasone, with or without Rituximab), CODOX-M/IVAC +/- R (cyclophosphamide, vincristine, doxorubicin, methotrexate, ifosfamide, etoposide, and cytarabine, with or without Rituximab), and dose-adjusted (DA) R-EPOCH (rituximab, etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin) (294). Although BLs and HLs are generally curable, these regimens for relapsed aggressive B cell lymphomas are typically ineffective. Currently a novel chemotherapeutic approach is under investigation in a phase II clinical trial (NCT01964755) for relapsed EBV-associated lymphomas that utilizes a combination of drugs to potentiate the function of zidovudine (ZDV) to suppress NF-κB and viral latency. Unlike in lymphomas, there is limited evidence on the clinical gains of chemotherapy alone in EBV+ epithelial cancers. Surgery (typically gastrectomy) and chemotherapy remain the first line of treatment for patients with EBV+ gastric cancer. Nevertheless, the efficacy of chemotherapy remains speculative. Corallo et al. reported that 6 EBV+ GC patients who received fluorouracil and platinum as first-line of chemotherapy had a 3-year survival rate of 80% compared to 26.5% in EBV– GC patients (295). However, another observational cohort study with 31 patients reported an overall response rate of only 29% in metastatic EBV-GC patients who received taxane/trastuzumab, fluoropyrimidine and platinum as the first-line therapy (296). Owing to the single center nature of these studies and small sample sizes, it is necessary to confirm these observations in larger cohorts and clinical trials (297). From the limited evidence it seems that patients with EBV+ GC have few metastases, longer survival, and high disease control rates. Although, chemotherapy helps some patients by increasing the frequency of event free and overall survival, it is still insufficient to treat EBV+ cancers completely and eradicate infected cells. Combining chemotherapy with immunotherapy has provided encouraging preliminary results but further exploration and development of more effective combinatorial strategies are required.
Adoptive cell therapy (ACT) is a form of immunotherapy where specialized in vitro expanded or modified immune (usually T) cells are transferred to patients to enhance or repress immunity. These T cells are either specific for an antigen (e.g. viral protein or tumor-associated antigen) or are genetically engineered to express chimeric antigen receptor (CAR) or modified T cell receptor (TCR). The first ACT was performed in 1994, where donor leukocytes which included EBV-specific cytotoxic T cells (CTLs) were infused into 5 patients who had developed EBV-associated PTLD. Complete remission was observed in 5/5 patients, however all of them developed graft-versus-host disease (GVHD) due to alloreactive T cells (298). Since then, significant progress has been made in the field of ACT for patients with EBV-associated PTLDs and the concept is expanding to include patients with NPC and HL, as summarized elsewhere (299). One of the strategies to minimize alloreactivity (i.e. to reduce the risk of GVHD) includes infusion of in vitro stimulated and expanded EBV-specific CTLs that are donor-derived (300, 301). Interestingly, adoptively transferred CTLs not only restore the anti-EBV immune response but can also establish long-term persistence (302, 303). Similar strategies for ACT have been developed for patients with NPC and HL, where CTLs specific to EBV latent antigens (e.g. EBNA1, LMP1, LMP2) are expanded ex vivo and infused into patients (304). Phase I/II clinical trials with such immunotherapy approaches have increased the overall survival of patient with recurrent or refractory NPC (305–307) and HL (308). A modified version of ACT for the treatment of EBV+ cancers is to engineer T cell receptors before transferring T cells to the patients. Such EBV-specific TCR-engineered T cell therapy is based on the rationale that TCRs on CD8+ T cells can be re-engineered to specifically recognize EBV latent and lytic proteins. The stability and anti-tumor effect of these chimeric TCRs have been evaluated in murine models and provide encouraging results (309). For instance, T cells expressing LMP1-specific TCR inhibited tumor growth and prolonged survival in xenograft mice (310). Another type of ACT being widely investigated is EBV-specific CAR T cell therapy. CAR-T cell therapies targeting specific antigens, for instance CD19, CD20, CD22 and CD30, have provided encouraging results for treatment of lymphoma in clinical trials (311–313). One drawback of CAR T cell therapy is that CD8+ T cells engineered with a CAR will also express their own native TCR, so the potential for auto-reactivity remains. Another the major limitation of CAR-T cell therapy is specificity for tumor-associated antigens, as some of these might be expressed by normal cells. This leads to adverse off-tumor toxicity, cytokine release syndrome and deficiencies in B-cell mediated humoral responses. Some of these limitations have been addressed by developing CAR-T cells that are specific to EBV antigens such as LMP1, since they would be specifically expressed in EBV infected malignant cells (314). In fact, LMP1-specific CAR-T cells exhibit enhanced tumor inhibition in LMP1-positive NPC xenograft mouse models (315). The translation of these therapies for the treatment of EBV-associated cancers warrants further evaluations before it can be prescribed as personalized immunotherapy for EBV+ cancers in humans.
Several antiviral drugs have been identified and are being currently evaluated for clinical use. These can be broadly divided into three classes: 1) nucleoside analogs such as acyclovir (ACV), ganciclovir (GCV), penciclovir (PCV), and their oral prodrugs valacyclovir (VACV), valganciclovir (VGCV) and famciclovir (FAM), respectively; 2) nucleotide analogs such as cidofovir (CDV); and 3) pyrophosphate analogs, including foscarnet. Although these antiviral agents have been clinically evaluated for different viruses, their clinical utility in the context of EBV-associated malignancies is lacking. To date, there is no effective Food and Drug Administration (FDA) or European Medicines Agency (EMA) approved antiviral therapy available for EBV infections. Nevertheless, we will briefly discuss some of these antiviral agents in the context of EBV-associated diseases. A more in-depth review of these drugs can be found published elsewhere (316). Nucleoside analogs such as ACV and GCV inhibit EBV in vitro. The antiviral effect of ACV is attributed to the preferential incorporation of its triphosphate into the viral DNA due to high-affinity interaction with EBV polymerase compared to the cellular polymerase. This process irreversibly and specifically terminates viral DNA elongation and replication (317). The effective dose of ACV against EBV is orders of magnitude lower (0.3μM) than host cells (250 μM) (318), resulting in a highly favorable therapeutic index and toxicity profile. The antiviral effect of GCV is greater than ACV but it is more toxic. Antiviral (e.g. ACV, GCV or VACV) prophylaxis has significantly reduced development of PTLD in high-risk EBV-seronegative lung transplant patients (319) and reduces EBV viremia in pediatric renal transplant patients (320). Clinical trials administering ACV along with prednisolone have shown the inhibition of EBV replication in the oral cavity, however they do not alleviate the duration or intensity of clinical symptoms (321). Importantly, none of the nucleoside analogs have any effect on latent infections. This is because the viral enzymes that are needed for the prodrug activity are not expressed during latent phase (322). Other nucleoside analogs with efficacy against varicella zoster virus, such as omaciclovir, have not yet been evaluated against EBV (323). Cidofovir is a nucleotide analog that possesses both antiviral and antiproliferative properties and is metabolized into its active form by cellular kinases (324). The antiproliferative effect of Cidofovir on EBV-infected NPCs has been previously reported (325). Consistently, intra-tumoral injection of cidofovir suppresses tumor growth in EBV+ NPC xenografts in nude mice (326). Another study showed that treatment of NPC (C15) and BL (Raji) cell lines with cidofovir decreases expression of LMP1 and EBNA2 oncoproteins and increases apoptosis and enhances ionizing radiation (IR)-induced regression of EBV+ NPC and BL tumor bearing nude mice (327). Tenofovir (TFV) is an acyclic nucleoside/nucleotide analog that has already been approved for the treatment of HIV and HBV infection, where it acts as an inhibitor of the viral reverse transcriptase (328). The prodrugs of tenofovir, disoproxil fumarate (TDF) and tenofovir alafenamide (TAF), are both orally bioavailable and are more potent than ACV, PCV and GCV (329). All of these antiviral agents target EBV DNA polymerase, however, unlike others, the tenofovir prodrugs are metabolized independently of viral enzymes to their active forms and depend on host enzymes, thus, permitting their usage for latently infected cells (330). Despite the availability of a plethora of antiviral agents, there aren’t any effective antivirals for EBV-associated cancers. However, studies on compounds like tenofovir holds some promise and warrants further study (331). Foscarnet is a pyrophosphate analogue with a broad antiviral activity against the Herpesviridae family (332). As a pyrophosphate analogue, it disrupts viral DNA polymerase activity by inhibiting cleavage of pyrophosphate from the nucleoside triphosphate. Unlike ACV and GVC, foscarnet does not depend on viral protein kinases for its activity, making it useful in cases of ACV/GCV acquired resistance. However, it might be less tolerated in patients due to increased toxicity. Several case reports have shown benefits of foscarnet in treatment of EBV+ PTLDs (333, 334). However, the systematic efficacy of foscarnet in treatments of EBV-associated needs to be further evaluated.
As discussed, latent EBV infection is associated with human malignancies, such as BL, PTLD, NPC, GC, HL and non-HLs. GCV and ACV are commonly used antiviral drugs that require the EBV lytic encoded protein kinase (EBV-PK) and thymidine kinase (EBV-TK) for the conversion of pro-drugs into active viral drugs. As a result, these drugs are inefficient in eliminating EBV-infected cells that are in the latent state. One therapeutic approach is therefore the induction of EBV lytic replication, also known as cytolytic virus activation (CLVA), in combination with antiviral drugs to enable specific targeting of tumor cells that harbor EBV in a lytic state (18, 335). CLVA within infected tumor cells can induce i) a cytotoxic or cytostatic effect from the lytic viral proteins; ii) expression of viral enzymes that metabolize and activate antiviral pro-drugs, such as ACV and GCV; and iii) a range of antigenic viral proteins that can now be recognized by host-immune cells (336) ( Figure 4 ). Although, several classes of lytic inducers have been identified and their mechanism of action has been elucidated in different cell types, only one clinical study has reported a promising outcome in a small fraction of patients with EBV+ tumors (337, 338). This is primarily because these compounds have three major drawbacks limiting their use in clinical settings. First, most of these compounds have low efficiency and can induce the lytic cycle in only a small percentage of cells, therefore a considerable proportion of cells are refractory (339). Second, the efficiency of these lytic inducers is heavily dependent on the cell type, thus cannot be broadly utilized for all EBV-associated malignancies. Many lytic inducers also have serious side effects, therefore their translation into clinical use is challenging. Lastly, there is a concern that chemical induction of EBV could promote viral dissemination (340, 341). Over the years, scientists have identified several classes of organic and chemical compounds that are able to induce the EBV lytic cycle in latently infected cells. Protein kinase C (PKC) activators (PMA), HDAC and DNA methyl transferase (DNMT) inhibitors, chemotherapeutic agents and anti-IgG are among the known classes of lytic inducers. Interestingly, evidence suggests that several distinct mechanisms of lytic induction may exist because synergistic effects have been observed when different lytic inducers are combined. For instance, treatment of BL and GC cell lines with combinations of PMA and sodium butyrate or valproic acid and cisplatin leads to significantly higher EBV lytic reactivation compared to individual treatments. Large-scale chemical library screens in GC and NPC cells have identified two additional distinct compounds that could induce EBV reactivation, one that resembles iron chelators and one that activates the MAPK pathway (342). Different classes of lytic inducers has been previously reviewed here (339). In this section, we will briefly discuss HDAC inhibitors and iron chelators that have been reported to induce the EBV lytic cycle via PKC-δ and HIF-1α pathways, respectively (343).
Of all the different inducers of the EBV lytic cycle, inhibitors of histone deacetylase (HDACi) have been well studied. These include sodium butyrate (NaB), valproic acid (VPA), suberoyl anilide hydroxamic acid (SAHA or Vorinostat), and romidepsin. Use of HDACi alone or in combination with GCV are currently being tested for use in patients. For example, butyrate in combination with GCV has shown promising results in patients with refractory EBV+ lymphoid malignancies (344). However, the efficacy of this combination therapy has received limited success in vivo due to the poor oral bioavailability and short half-life of butyrate (345). A systematic study of HDACi have ranked panobinostat, belinostat, butyrate, entinostat, oxamflatin, apicidin, and largazole from highest to lowest in their ability to induce EBV-TK and EBV-PK kinases, suggesting that despite the structural diversity, most HDACi can function as inducers of EBV lytic replication (346). Interestingly, despite belonging to the same class, these HDACi might invoke different mechanisms of action to induce EBV reactivation. For instance, VPA antagonizes the ability of other HDACi to induce EBV lytic reactivation (347).
Iron is a nutrient that plays an important role is various aspects of cell biology including growth and differentiation. It is used by heme-containing proteins and serves as a cofactor for many enzymatic activities. Desferioxamine is a widely used drug to treat iron-overload. Desferrithiocin, an orally available iron chelator, is a more potent substitute of Desferioxamine due to higher bioavailability. Both these compounds adversely inhibit proliferation of T cells, which can be rescued by the addition of iron (in the form of ferrous chloride, FeCl2) (348). Iron chelators induce the EBV lytic cycle in certain cancer cells by inhibiting enzymatic activity of protein hydroxylases (349). A novel compound, named C7, has been recently described and reported to induce early, but not late, lytic proteins via intracellular iron chelation, alleviating the concern about viral dissemination (342, 350). However, C7 seems to direct only a small proportion of cells into the lytic cycle. Moreover, due to the abundance of iron in vivo and within the TME, delivery of iron chelators into tumors remains challenging, suggesting that further studies are needed to identify novel compounds and/or overcome these limitations.
Due to highly penetrant effects, small molecule inhibitors have been of great interest in treating various cancers. In viral-associated cancers, essential viral proteins are great targets because inhibitors against them might have lower toxicity against host cells. In EBV-associated cancers, since EBNA1 has served as a promising target for virus-targeted therapies as it has no cellular homologue and is constitutively expressed in all EBV+ cancer cells. There are several ways by which EBNA1 can be functionally perturbed, and significant advances are currently being made in the field of EBNA1 targeting therapies. These include inhibiting DNA binding activity of EBNA1 (351, 352), disrupting homodimerization (353), blocking its interaction with key host cellular proteins, such as USP7, CK2 or targeting its oncogenic partners, such as MDM2 (354). VK-2019 is a small molecule inhibitor that binds to EBNA1 and disrupts its DNA binding activity. It is administered orally and is currently undergoing phase 2 clinical trial (NCT03682055) for patients with advanced NPC. Further studies are needed to identify specific inhibitors of other latent EBV proteins, including LMP1, or viral genes that are essential for EBV transformation.
To our knowledge, there are currently no FDA-approved vaccines for EBV. Since EBV is a causative agent for a range of diseases a prophylactic or preventative vaccine would be the most beneficial and cost-effective therapeutic approach to manage EBV-associated IM, malignancies and autoimmune diseases. The rationale for prophylactic vaccination is to prevent EBV infecting its target cells by inducing an antibody response. Below we will briefly discuss strategies for developing prophylactic vaccines against EBV. Considering that EBV requires multiple envelope proteins to enter target cells, these serve as excellent candidates for developing recombinant envelope protein vaccines. EBV glycoprotein gp350 (BLLF1), is the most abundant envelope protein, and initial studies on EBV vaccine development primarily focused on gp350 (355, 356). However, a large-scale clinical trial using soluble gp350 failed in preventing infection, albeit reducing the development of IM after EBV infection (357). In addition, gp350 vaccine candidates only protect B cells but no other EBV target cells (e.g. epithelial cells) from infection. EBV glycoproteins gH/gL and gp42 bind to HLA-DR on B cells and integrins and ephrin receptor A2 on epithelial cells, respectively, and facilitate EBV fusion to the cell membrane (70, 358–360). Since, these viral glycoproteins are integral components of the core viral fusion machinery, they also serve as excellent candidates for prophylactic vaccine development (299). To this end, in 2019, Bu et al. developed an EBV gH/gL/gp42 based nanoparticle vaccine. This vaccine inhibits EBV infection of both epithelial and B cells by eliciting an antibody response that targets the virus membrane-fusion proteins in mice and non-human primates-macaques (356). This year, a novel EBV gp350-ferritin nanoparticle vaccine was developed by researchers in NIAID/NIH that has entered the Phase I clinical trial to determine its safety and immunogenicity in humans (NCT04645147). Taken together, these studies suggest that targeting multiple EBV glycoproteins - gH/gL, gB and gp350 - together could synergistically induce highly effective EBV neutralizing activity. Additionally, evidence suggests that these viral glycoproteins can also induce T cell immune responses to further enhance vaccine efficacy by recruiting T cells to either kill or inhibit transformation of recently infected cells if neutralizing antibodies are ineffective, for example due to variations in EBV protein sequences (361, 362). Recombinant viral vectors are also commonly used to develop therapeutic vaccines. Essentially, these are live viruses that are engineered to express specific proteins that help elicit an immune response. Such vaccines can infect target cells and induce a CD8+ T cell response, enhance the anti-inflammatory response by serving as adjuvants themselves and have high gene transduction efficiency (363). The first EBV vaccine was developed in 1995 and tested in humans. This was a live recombinant vaccinia-based virus, expressing EBV envelope protein BLLF1/gp350 (364). However, this vaccine was discontinued due to adverse effects. In 2004, Taylor et al. developed a chimeric antigen construct using a modified vaccinia virus “Ankara” (MVA) vector that encoded the C-terminal portion of EBNA1 and entire LMP2 (MVA-EL). Upon transduction, the EL protein can be processed by HLA I and II, resulting in CD8+ and CD4+ T cell responses (365). Since these two EBV latent proteins are expressed in NPC, MVA-EL was tested for safety and immunogenicity as a therapeutic vaccine for patients with EBV+ NPCs in phase I clinical trials (NCT01147991). Indeed, this vaccine was well tolerated and induced EBV-antigen specific T cell responses in 8/14 patients in UK and 15/18 patients in Hong Kong (366). Further studies are needed to determine its translation to the clinical setting for treatment, either alone or in combination with other modalities including T cell therapies. In 2012, another group developed and evaluated the ability of a recombinant adenoviral vector-based vaccine (AdE1-LMPpoly) to induce EBV-specific T cell responses in recurrent or metastatic NPC in a phase I clinical trial (ACTRN12609000675224). Encouragingly, EBV-specific T cells were expanded in 16/24 NPC patients and infusion of AdE1-LMPpoly–generated T cells was tolerated and prolonged survival by 2.3-fold. A phase II randomized clinical trial is necessary to confirm these observations (367). A comprehensive list of current vaccines is available here (368, 369). In addition to the EBV envelope protein-based vaccine and recombinant viral vectors, development of viral like particle (VLP) vaccines is another area of active research. The design of such viral particles is based on the rationale that a non-infectious version of EBV will elicit an EBV-specific innate and adaptive immune response in a safe and effective manner (370). Mechanistically, VLPs are phagocytosed and processed by DCs. DCs then activate CD8+ and CD4+ T cells by presenting viral antigens on HLA class I and II, respectively (371). While CD8+ T cells mount a cytotoxic response, CD4+ T cells elicit an anti-tumor response via Th1 and Th2 type responses that produce pro-inflammatory cytokines (e.g. IFN-γ, TNF-α, and IL-4, IL-10, respectively) (371). The status of VLP developments against oncoviruses and their biological and chemical characterization has been recently reviewed (372). In 2015, a novel EBV vaccine based on the Newcastle disease virus (NDV) VLP platform was developed, consisting of EBVgp350/220 ectodomain fusion protein that structurally mimicked EBV. This VLP elicited a long-lasting neutralizing antibody response in mice, but the responses were comparable to soluble gp350/220 (373). Thus, a more immunogenic VLP was developed that incorporated additional EBV glycoproteins and latent antigens – EBNA1 and LMP2. Immunization with gH/gL-EBNA1 and gB/LMP2 VLPs produced high neutralizing antibody titers in vitro and EBV-specific T cell responses in vaccinated BALB/c mice (374). DNA-free VLPs/LPs typically consists of EBV structural proteins that are weakly immunogenic towards CD8+ T cells. As a result, humoral and cell-mediated immune responses that recognize these structural proteins offer limited to no protection against latently infected cells. Another strategy is to use EBV particles themselves for VLP vaccines. The first EBV VLP was created by removing the terminal repeats that result in production of large amounts of defective viral particles without the viral DNA that could bind to both B and epithelial cells (375, 376). In subsequent studies, more viral packaging proteins (BFLF1, BFRF1, BBRF1) and viral oncogenes (EBNA2, 3A, 3B and 3C, LMP1 and BZLF1) were deleted to improve the safety profile of these VLPs, while maintaining immunogenic potential (377, 378). In 2018, a more immunogenic EBV VLP was created by fusing EBNA1 and EBNA3C to the EBV tegument protein BNRF1. As a result, only 14% of mice vaccinated with modified VLPs had detectable viral load in the peripheral blood compared to 100% of the control PBS-vaccinated mice (379). While, VLP-based therapeutics are being developed and evaluated in early clinical trials, none have yet reached the phase III efficacy clinical trial stage. This is attributed to two major limitations. First, they suffer from production efficiency and scalability, which is partly because of the use of mammalian cells that lead to low viral titers and the presence of contaminants from human producer cell lines. Second, they offer low immunogenicity as epitope-based vaccines. This necessitates the need to co-administer an adjuvant or design better antigen delivery systems. Nonetheless, there are some promising ongoing studies on EBV-derived VLP vaccines to prevent EBV+ cancers. With the success of mRNA vaccines against SARS-CoV2, researchers have developed an EBV mRNA vaccine based on the same platform. Moderna has recently launched phase I clinical trial of its EBV mRNA vaccine (mRNA-1189) that have shown high EBV neutralizing antibody titers in mice (NCT05164094). This investigational vaccine targets EBV glycoproteins - gp350, gB, gH/gL and gp42 - and is hypothesized to prevent IM and EBV infection. Despite the potential challenges with mRNA-based approach for an asymptomatic virus like EBV (299), results from these trials will help in developing effective preventative treatment approaches for EBV infections and related disorders.
There are several malignancies that associated with EBV infection. Despite decades of research in this field, the precise role of this virus in the tumorigenic process and immunoevasion is not fully understood. While EBV+ cell lines, including LCLs, serve as a good model system to study EBV-host biology in vitro, mouse models help investigate and understand EBV-specific biology in vivo. Since the γ-herpesvirus have co-evolved along with their hosts, i.e. humans and monkeys, there is a lack of suitable counterparts in rodents (380). Unfortunately, the murine γ-herpesvirus 68 (MHV-68) lacks EBV’s transforming ability and thus fails to recapitulate EBV-induced tumorigenesis (380). Consistently, major differences in molecular mechanisms, tumorigenesis, cellular tropism, and immune responses between MHV-68 and EBV infections have been observed (381). As such, murine models used to study EBV-associated malignancies are typically immunodeficient. Transferring human peripheral blood mononuclear cells (PBMCs) from EBV-seropositive human donors into immunodeficient mice can generate PBMC-derived EBV+ B cell tumors (382). Zhang et al. further developed a genetically engineered mouse model to study EBV-driven lymphomas found in immunosuppressed patients (383), underscoring the relevance of immunodeficient murine models for the study of EBV disease (382). However, since these mice are immunocompromised, they are unable to induce a host immune response. Transferring human PBMCs to overcome this barrier often leads to severe xeno-graft versus host disease (GVHD) in these mice, limiting the duration of the study (384). Recognizing these limitations, the field has shifted to developing and using lymphocyte-deficient mice which have the potential to be reconstituted with human immune components (385). These mice are valuable resources to investigate human specific viral infections, as well as develop and test therapeutic vaccines (386). A more comprehensive review of murine models for EBV-associated tumorigenesis can be found elsewhere (386, 387). Despite being an excellent model system for EBV-associated lymphomas, these mice are not suitable to study EBV+ epithelial cancers since the mice lack human epithelial cells. In these cases, nonhuman primate animal models might be of justified use. These have been reviewed by others (388).
With the advent of massively parallel sequencing technologies, researchers have been able to better characterize the complexity of interactions between host and viral genes and identify novel genomic and epigenomic alterations within EBV-infected cancer cells that are druggable. Specifically, these technologies enable the capture of all the genetic material inside host cells, which can then be used to simultaneously study the biology of both host cells and infecting viruses (180, 389–392). These studies specifically overcome a limitation of laborious traditional EBV genetic studies where only individual viral or host genes are studied in isolation outside of the tumor context. In this section, we will discuss some of the sequencing approaches that are used to explore the genetic, transcriptomic and epigenomic landscapes in the context of EBV-associated malignancies.
The investigation of EBV genome sequences is important due to their association with several human malignancies (165). Prior to 2013, GenBank had whole genome sequences from less than 10 strains of EBV (393). Conventional sequencing techniques were inefficient, expensive and could help investigate only a few viral genes at a time. They typically involved digestion of genomic DNA via restriction enzymes, cloning and Sanger sequencing. Moreover, the large size of the EBV genome (~172kbp) added to the cost and time. Whole genome sequencing (WGS) technologies marked a new era of EBV genome sequencing and could be performed with or without EBV enrichment (180, 394). The WGS approach is quite sensitive to detecting viral derived sequences as well as integrated viral regions within the host genomes. As such, WGS has not only helped identify distinct EBV variants, mutations and oncogenes but has also allowed for a comprehensive survey of EBV integration in a wide variety of human malignancies as reviewed here (395, 396). Additionally, WGS of EBV-associated tumors have been extremely informative in identifying a range of distinct host variations that promote cancer (65, 186, 214, 217).
The use of high-throughput sequencing technologies to study the transcriptomic landscapes of disease have become a common practice. The general workflow for such techniques begins with bulk RNA extraction from the biospecimens under investigation, followed by RNA selection (mRNA or ribosomal-free RNA), cDNA synthesis, library preparation and sequencing. In the case of viral-associated malignancies, such techniques have provided key insights into cellular genes and pathways that are affected by virus (397, 398). Additionally, since these sequencing technologies are agnostic of the origin of the RNA within the cells, they capture genetic materials of both host and infecting viruses (391, 392, 399). As such, transcriptomics studies of EBV-associated malignancies have revealed important aspects of EBV biology including expression program and its interactions with the host to promote disease (180, 390, 400, 401). For example, a large-scale transcriptomic study of EBV-associated cancers classified EBV+ cancer types into molecular sub-types according to activation or repression of interferon signatures which was correlated with expression of several immune checkpoint genes such as PD-L1 and IDO1 (180). Of note, the heterogeneity of cells within the biospecimens, specifically tumors and their microenvironments, present a challenge for interpreting bulk transcriptomics data. For example, a change in expression of a gene or activation of a pathway could reflect either a change in tumor cells and/or a change in the cellularity of the TME, for example by immune cell infiltration. Recent computational tools that can deconvolute the composition of cells from bulk data have provided some remedy for this issue (402–404). Nevertheless, the accuracy of these tools remains limited, and they have not yet widely applied to study TME in EBV-associated malignancies. The advent of single-cell RNA-sequencing (scRNA-seq) has overcome this challenge and has revolutionized the field of transcriptomics and helped scientists map and generate individual cell atlases (405). Specifically, scRNA-seq has enabled the study of host-pathogen interaction in viral-associated diseases as well as cellular heterogeneity (406, 407). scRNA-seq is rapidly being adopted to study EBV-associated malignancies. For example, recent studies have revealed the landscape of both tumor and infiltrating immune cells in NPCs that are associated with prognosis (234, 235, 408). Additionally, scRNA-seq are now used to delineate rare cell subpopulations such as cancer stem cells, cell-cell interactions via receptor-ligand analyses, cell differentiation via trajectory and time-resolution analyses (409). Additional steps also allow TCR, BCR and cell-surface protein expression (CITE-seq) sequencings at the single cell level to supplement scRNA-seq for the study of the diversity of immune cells. Given these technologies, it is exciting to get detailed understanding of how EBV affects these aspects of biology across various diseases and conditions, such as response to therapeutic treatments. It is important to point out that the detection of lowly expressed genes in scRNA-seq is challenging and faces a frequent “drop-out” where it might be only sporadically detected across different cells. This could be an issue for detecting EBV genes in malignancies associated with latency where most EBV genes are expressed at low levels.
Transcriptional regulation, epigenetic changes and physical interactions are paramount to EBV biology and understanding EBV-associated diseases. Appending high-throughput sequencing to traditional lab techniques such as chromatin immunoprecipitation (ChIP) has enabled genome-wide detection of transcription factor (TF) bindings and epigenetic changes, such as histone modifications and DNA methylations. For instance, ChIP-seq of EBV TFs has revealed many binding sites across the host genome, which has implications for pathogenic mechanisms (410). Conversely, many cellular TFs can also bind the EBV genome (390). ChIP-seq analysis of EBV infected GC and NPC cell lines reveal a redistribution of characteristic histone marks such as H3K4me1/3 and H3K27ac (411). Of note, one of the limitations of these technologies is their requirement for millions of cells, however, recent technologies such as CUT&RUN/Tag sequencing have resolved this issue. Additional adjustments to high-throughput sequencing technologies by ligating proximal chromatin regions (e.g., Hi-C) has enabled to study physical interactions between genomic loci. This technology has enabled the study of interactions between EBV episomes and host genomes that are consequential to host gene regulation (412, 413). There has been a large number of additional assays deployed to study specific aspects of EBV biology genome wide (414, 415). For example, assay for transposase-accessible chromatin using sequencing (ATAC-seq) has been developed to study chromatin accessibility and has been extensively used to study how EBV infection affects chromatin accessibility. It is important to note that most of these assays utilize bulk sample processing and therefore observations in heterogeneous cell populations should be carefully interpreted, as discussed above for bulk RNA-seq. Recently, the limits of some of these technologies have been pushed to the single cell level and integrated into scRNA-seq platforms to enable multiomics based investigation of EBV infection (409).
As the above studies demonstrate, EBV is a causative agent and/or associated with a plethora of diseases including cancer and autoimmunity. Despite decades of research, the underlying mechanisms governing how the interactions between EBV and host cells promote carcinogenesis are incompletely defined. As a result, effective and individualized treatments for EBV-associated diseases still remain either non-specific or lacking. Nevertheless, it is also obvious that EBV is heavily regulated by a variety of factors, and it extensively regulates cellular processes and the microenvironment. High-throughput methods are ideal for revealing complex networks of tissue and disease associations. As discussed, these technologies have their own limitations. Utilizing orthogonal and multiomics technologies can typically overcome some of these limitations. For example, one of the limitations of single-cell transcriptomics is the loss of spatial information during the tissue dissociation, which are important for understanding disease biology. Recent high resolution (down to the sub-cellular level) spatial transcriptomics methods can help overcome such issues, however, due to their recent development, they have not yet been employed to study EBV-associated diseases. Additional factors such as the microbiome might also be relevant to EBV-disease biology (416) and should be considered in designing tools and models. Rigorous computational modeling is also needed to accurately identify shared or tissue-specific signatures across EBV-associated diseases. Lastly, as discussed above, better animal models and drug delivery systems are needed in order to translate laboratory findings.
All authors contributed to the article and approved the submitted version.
This work was supported by extramural research programs of the NIH (R35GM138283) and the Showalter Trust (research award to MK). This research was supported (in part) by the Intramural Research Programs of the National Institute of Diabetes and Digestive and Kidney Diseases (project number ZIA/DK075149 to BA).
The authors also gratefully acknowledge the SIRG Graduate Research Assistantships Award to SC and support from the Purdue University Center for Cancer Research, P30CA023168.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647128 | Amrendra Kumar,Kanak Raj Kanak,Annamalai Arunachalam,Regina Sharmila Dass,P. T. V. Lakshmi | Comparative transcriptome profiling and weighted gene co-expression network analysis to identify core genes in maize (Zea mays L.) silks infected by multiple fungi | 27-10-2022 | transcriptome analysis,WGCNA,CAZymes,R-gene,transcription factor,fungal,maize silks | Maize (Zea mays L.) is the third most popular Poaceae crop after wheat and rice and used in feed and pharmaceutical sectors. The maize silk contains bioactive components explored by traditional Chinese herbal medicine for various pharmacological activities. However, Fusarium graminearum, Fusarium verticillioides, Trichoderma atroviride, and Ustilago maydis can infect the maize, produce mycotoxins, hamper the quantity and quality of silk production, and further harm the primary consumer’s health. However, the defense mechanism is not fully understood in multiple fungal infections in the silk of Z. mays. In this study, we applied bioinformatics approaches to use the publicly available transcriptome data of Z. mays silk affected by multiple fungal flora to identify core genes involved in combatting disease response. Differentially expressed genes (DEGs) were identified among intra- and inter-transcriptome data sets of control versus infected Z. mays silks. Upon further comparison between up- and downregulated genes within the control of datasets, 4,519 upregulated and 5,125 downregulated genes were found. The DEGs have been compared with genes in the modules of weighted gene co-expression network analysis to relevant specific traits towards identifying core genes. The expression pattern of transcription factors, carbohydrate-active enzymes (CAZyme), and resistance genes was analyzed. The present investigation is supportive of our findings that the gene ontology, immunity stimulus, and resistance genes are upregulated, but physical and metabolic processes such as cell wall organizations and pectin synthesis were downregulated respectively. Our results are indicative that terpene synthase TPS6 and TPS11 are involved in the defense mechanism against fungal infections in maize silk. | Comparative transcriptome profiling and weighted gene co-expression network analysis to identify core genes in maize (Zea mays L.) silks infected by multiple fungi
Maize (Zea mays L.) is the third most popular Poaceae crop after wheat and rice and used in feed and pharmaceutical sectors. The maize silk contains bioactive components explored by traditional Chinese herbal medicine for various pharmacological activities. However, Fusarium graminearum, Fusarium verticillioides, Trichoderma atroviride, and Ustilago maydis can infect the maize, produce mycotoxins, hamper the quantity and quality of silk production, and further harm the primary consumer’s health. However, the defense mechanism is not fully understood in multiple fungal infections in the silk of Z. mays. In this study, we applied bioinformatics approaches to use the publicly available transcriptome data of Z. mays silk affected by multiple fungal flora to identify core genes involved in combatting disease response. Differentially expressed genes (DEGs) were identified among intra- and inter-transcriptome data sets of control versus infected Z. mays silks. Upon further comparison between up- and downregulated genes within the control of datasets, 4,519 upregulated and 5,125 downregulated genes were found. The DEGs have been compared with genes in the modules of weighted gene co-expression network analysis to relevant specific traits towards identifying core genes. The expression pattern of transcription factors, carbohydrate-active enzymes (CAZyme), and resistance genes was analyzed. The present investigation is supportive of our findings that the gene ontology, immunity stimulus, and resistance genes are upregulated, but physical and metabolic processes such as cell wall organizations and pectin synthesis were downregulated respectively. Our results are indicative that terpene synthase TPS6 and TPS11 are involved in the defense mechanism against fungal infections in maize silk.
Maize (Zea mays L.), also called the “queen of cereals”, ranks third in the world after wheat and rice production. About 5.5% of maize (corn) is used as human food from all energy sources of food (51%), which comes from rice (20%), wheat (20%), and other cereals or grains (6%). It is also one of the most widely grown grain crop and is being cultivated in more than 166 countries. The United States produces most of the maize (30%), followed by China (23%), Brazil (9%), Argentina (5%), and India (2%) (Crop et al., 2021). Maize is primarily grown for food and feed intent for human and animal nutrition. In addition, maize has found extensive applications in beauty and drug industries, too. All plant parts in Z. mays can be used to generate revenue. The silk from Z. mays has been used to treat different illnesses as it is being applied in the Indian system of medicine and Chinese traditional medicine (Zhao et al., 2012). In fact, in India, between 2,500 BCE and 500 BCE, the ayurvedic concept saw Z. mays as an essential herb, especially the silk part (stigma maydis), for healing and controlling many diseases (Pandey et al., 2013). This traditional knowledge of the significance of the use of maize silk was eventually lost with the advent of allopathy in the 18th century. Recent research has shown that maize silk exhibits powerful health-promoting effects. This is also true because the silk contains bioactive compounds such as flavonoids, proteins, carbohydrates, vitamins, steroids, tannins, alkaloids, mineral salts, and polysaccharides (Zhao et al., 2012; Guo et al., 2017). These compounds may help protect against cancer, hypertension, diabetes, hepatic, cardiovascular, and other age-related diseases. Researchers are exploring ways to lower body weight and blood glucose levels, increase serum insulin secretion, improve glucose intolerance in type 2 diabetic mice, and control hyperglycemia (Mada et al., 2020). Rahman and Wan Rosli (2014) opined that maize silk provided an ideal environment for fungal propagules as a nutrient-rich, soft, and moisture-laden tissue within the husks. The corn silk serves as an ideal place for fungal propagules to reside and multiply within the cob environment. The fungal spores adhere and germinate into hyphal structures, which spreads into the maize silk, infects the ovules, and creates an imbalance of the hormones (Li et al., 2018). The parenchymatous cells of the maize silk serves as a suitable place for hyphae to grow especially for fungi like Aspergillus, Fusarium, Penicillium, and Ustilago species which cause diseases like ear rot, corn smut, and brown spot (Miller et al., 2007) and an economic loss of 5–42% yield per year (Thompson and Raizada, 2018). The silk of Z. mays is also said to have an effect against Trichoderma species (Gong et al., 2014; Contreras-Cornejo et al., 2016). A study in 2019 reported that maize silk has the genes and transcription factors that code for the callose of the papillae, which prevent fungi from growing (Shi et al., 2019) within the cobs. Fusarium species like Fusarium graminearum (Fg) and Fusarium verticillioides (Fv) usually infect the outer layer of the maize silk in Z. mays in order to draw nourishment for hyphal growth. On the other hand, infection caused by Ustilago maydis was found to affect the entire length of maize silk. These fungi produce mycotoxins, carcinogenic substances that cause esophageal and liver inflammation in humans (Marín et al., 2004). It is therefore important to understand the mechanism of the plant–fungal interaction in the infection process. Transcriptome studies have been extensively used to study specific genes expressed during the infection process. Hence, multiple fungal systems from the families Nectriaceae (F. verticillioides—Fv and F. graminearum—Fg), Hypocreaceae (Trichoderma atroviride—Ta), and Ustilaginaceae (U. maydis—Um) were used to study the expression pattern in Z. mays silk. Agostini et al. (2019) chose the datasets from the experimental studies on the combination of these fungi to examine and figure out the essential genes involved in many molecular and biological processes. Furthermore, we looked at the co-expression in different networks using weighted gene co-expression network analysis (WGCNA) to build the networks based on the pairwise co-expression between gene expression levels. Since WGCNA builds a scale-free network based on similarities in gene expression profiles that may be linked to the phenotypes of interest, this method was used to find groups of genes that work well together (Bakhtiarizadeh et al., 2018; Abbassi-Daloii et al., 2020; Xu et al., 2021). The goal of the comparative study based on statistical estimates like the DEG and WGCNA modules was to investigate and study more about how different fungi infect Z. mays silk. F. graminearum, F. verticillioides, and U. maydis are all partially biotrophic parasites that can also eat dead organisms (Incremona et al., 2014; Pei et al., 2019; Pandian et al., 2020). F. verticillioides is an endophyte which competes with the fungal pathogen F. graminearum and is antagonistic to U. maydis (Lee et al., 2009; Rodriguez Estrada et al., 2012). This study evaluates how fungal stress resistance and yield can improve maize silk through molecular breeding and biotechnology.
Two transcriptome datasets of fungus-infected silk of Z. mays were obtained from the National Centre for Biotechnology Information-Sequence Read Archive (NCBI SRA) database. Each dataset consisted of samples (18) infected by different fungi belonging to the families of Hypocreaceae, Nectriaceae, and Ustilaginaceae, with BioProject accession numbers PRJNA13048 (A) (https://www.ncbi.nlm.nih.gov/bioproject/PRJEB13048) and PRJNA382306 (B) (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA362306). Each data set had three biological replicates comprising of silk samples affected with two fungi, F. graminearum and U. maydis in A, while B was infected with F. verticillioides and T. atroviride along with the control in each dataset (Agostini et al., 2019) ( Supplementary Table S1A ). The quality of both datasets was computed using the FASTQC tool (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/), and the raw read sequences of both datasets were mapped to the latest reference sequence of Z. mays B73 (V5.0) (http://ftp.ebi.ac.uk/ensemblgenomes/pub/release-51/plants/fasta/zea_mays) using the HISAT2 tool (Kim et al., 2015). The read count of each gene mapped to the reference genome was calculated using the FeatureCount tool (Liao et al., 2014).
Pairwise differential gene expression analysis was performed using control A and B datasets based on the experimental design ( Supplementary Table S1B ). DESeq2 of Bioconductor R package (Love et al., 2014) was used to perform differential expression calculation, and significantly differentially expressed genes (DEGs) were identified by applying the cutoff value of log2 fold change of ≥|1.5| with P-value cutoff <0.05. The expected significant common DEGs between control A and B datasets were filtered out for further analysis.
Weighted gene co-expression network analysis (WGCNA) (Langfelder and Horvath, 2008) of the R package was performed on the complete read count matrix of common DEGs identified (12,447 genes across 18 samples). The matrix between each pair of genes across all the samples was calculated using Pearson’s correlation. It generated an adjacency matrix by default soft power and computed the topological overlap matrix (TOM) along with the corresponding dissimilarity (1-TOM) values. Gene modules were detected using the dynamic cutting algorithm with a minimum module size of 30 and a default cutoff height (0.99), and the gene modules were arranged in the dendrogram depending on their shape (Langfeldera et al., 2007)
The correlation between module eigengenes and the gene expression of genes related to biotic stress was analyzed among the significantly correlated modules of interest associated with biotic stress in Z. mays silk. A heat map was used to represent the correlation values. The module membership (MM) is the association between each module eigengene and its gene expression as gene significance (GS), defined as the correlation between each trait and its gene expression (Tian et al., 2021). The MM and GS were determined to closely correlate the genes in a module with a cutoff value of MM >0.8 and a GS >|0.2| (Farhadian et al., 2021). The modules were correlated with the most significant DEGs common among relevant specific traits. All modules were considered core genes.
A protein–protein interaction (PPI) network was constructed for the more significant common gene accession using the STRING database (v11) (Szklarczyk et al., 2019) by applying an interaction filter score >0.4 (medium confidence) and further visualized through Cytoscape (Doncheva et al., 2019). Within the networks, the parameters K-mean cluster score = 2, degree cutoff = 2, and maximum depth = 100 (Bandettini et al., 2012) were set. A subnetwork analysis was performed using molecular complex detection (MCODE) for the clustering connected. For the identification of key/essential genes from the PPI network, the CytoHubba (Chin et al., 2014) plugin of Cytoscape was used, which extracted the top 100 genes with the selected four scoring methods of CytoHubba, namely, maximal clique centrality (MCC), maximum neighborhood component (MNC), edge percolated component (EPC), and node–connect degree, respectively.
To recognize and identify the functions of the significantly expressed genes, Biomart, ShinyGo, and UniProt (Kinsella et al., 2011; Consortium, 2015; Ge et al., 2020) were employed. The information was generated and further verified through BLASTx (Camacho et al., 2009) against PlantTFdb (Guo et al., 2007) and PRGdb version 3.0 (Osuna-Cruz et al., 2018) for the recognition of genes encoding for resistance, transcription factors, and specific pathways related to them. The DRAGO (PRGdb tool) pipeline also classified the R-gene classes and domain. In contrast, dbCAN2 with default parameters integrated with automated annotation tools of Diamond, HMMER, and Hotpep (Finn et al., 2011; Buchfink et al., 2015; Osuna-Cruz et al., 2018) enabled us to identify the polysaccharide degradation enzymes, also known as CAZymes.
To acquire a better understanding of the essential genes that are involved in the defense mechanisms in Z. mays silk against different fungal infections, a comparative study of the publicly accessible transcriptome information was carried out. The current study used DEG and WGCNA analyses to identify the core set of genes modulated in infected maize datasets. The core genes were further subjected to PPI for essential gene indication, followed by examining biotic stress determinants like TF, R-genes, and CAZymes. Supplementary Figure S1A provides a schematic of the analytical process. Supplementary Figure S1B displays the alignment data for each sample, and Supplementary Figure S1C illustrates the process by which readings were allocated to the genic feature count. The FastQC statistics showed that all the samples were of good quality, with above 80% of the samples lined up with the reference genome, and, further, that more than 70% of the reads were uniquely assigned to genic features.
The two datasets were analyzed for differential expression using DESeq2, and genes with a p-value of 0.05 and |log2FC|>1.5 were determined to be significant DEGs. Approximately 33% of genes (14,694 and 14,808 of 44,303 genes) of datasets PRJEB13048 and PRJNA362306 ( Supplementary Table S2 ) were differently expressed in pairwise differential expression analysis of fungus-infected versus control samples ( Table 1 ) ( Supplementary Figure S1D1–4 ). In the DEGs, approximately 28% (12,447) of genes appeared in both datasets ( Supplementary Figure S1E ). The number of DEGs varied in the fungal infections based on the comparison between up- and downregulated genes within controls. A comparative analysis of the DEG infections exhibited about 9,558 (22%) genes being expressed uniquely, among which roughly 5,043 (11%) genes were recognized to have been duplicated. Figure 1 displays that 19.4% (8,577) of Z. mays silk genes were significantly expressed in Fg, 11.7% (5,146) were described in Um, 1.7% (723) were expressed in Fv, and 0.3% (155) were expressed in Ta infections, respectively ( Supplementary Table S3 ). It is thus important to investigate which gene among them is responsible in regulating the expression under different fungal perturbations. The rationality variation in the value of the DEGs seemed to be caused by fungal infection (Yang et al., 2022), and this perhaps could be reasoned out with the differences observed in the sample sizes of the datasets, where the control of dataset A (Ca) had 7.5 GB, while the control of dataset B (Cb) had 2.5 GB. The concept well defends this examination that varied sample sizes and the spatiotemporal specificity of samples could influence the drastic variations in the different data sets (Wang, et al., 2021). Furthermore, a Venn diagram of total DEGs comparison between up- and downregulated genes within controls was examined in the interaction analysis of DEGs expressed in different fungal infections. In total, 21 (0.2%) genes were represented across the four different fungal infection settings. In the intersection of three fungal infection situations and specifically in Fg, Fv, and Ta infections, four genes (0.0%) were expressed. Additionally, in Fg, Ta, and Um infections, 17 genes (0.2%) were expressed. Fg, Fv, and Um infections expressed 479 (5.0%) genes. Two genes (0.0%) were expressed in Fv, Ta, and Um. The intersection between two fungal infection conditions, namely, Ta and Um, expressed 10 genes (0.1%). In Fv and Ta, 13 genes were expressed. Twenty genes (0.2%) were expressed under Fv and Um infection circumstances. Only Fg and Um expressed 3,838 (40.2%) genes. In the presence of both Fg and Fv infections, 77 (0.8%) genes were expressed. In Fg and Ta, 21 (0.2%) genes were expressed. Upon infection with a single fungal isolate of Fg, 4,120 (43%) genes were expressed, while 760 (8%) genes were expressed during the Um infection, and 108 (1.1%) and 68 (0.7%) genes were expressed in Fv and Ta infections, respectively ( Figure 2 ). These genes involved different physiological activities in Z. mays, including hormone response, secondary metabolism, phosphorylation, photosynthesis, cell wall organization, control, replication, and response to stimuli. Therefore, our basic premise is that these genes might have different gene expression patterns and counts depending on their natural genetic makeup. While some processes overlapped in samples of other fungal infections, Fg infection displayed higher gene expression levels in the Z. mays silk.
Weighted gene co-expression network analysis of 12,447 common genes was carried out to identify the core genes involved in fungus defense. The WGCNA analysis resulted in 11 modules, with 58 genes in the small modules to 5,415 genes in the most significant modules ( Figure 3A ). A correlation of module eigengenes to disease trait data was performed, with a cutoff value of significance |GS| >0.5 and P-value <0.05. Out of 11 modules, purple, yellow, and green-yellow modules were associated with both Ca and Ta, the green module with Ca, red and turquoise modules with Fg, black with Cb, pink being positively correlated with Um, and magenta, blue, brown, and black modules being negatively correlated with Fg ( Figure 3B ). Interestingly, none of the modules was significantly associated with Fv. A further intra-modular analysis based on the gene significance (GS) and module membership (MM) of genes identified vital genes in the six modules for the Fg trait. A filter of |MM >0.8 and GS >|0.2| was applied for essential gene identifications. The up- and downregulated genes were compared to the genes in modules with relevant specific traits (Yang et al., 2020). We were able to infer that the up- and downregulated core genes in modules 3,817 (607 and 3,210, red and turquoise modules) and 2,851 (1,949, 564, 212, and 126; blue, brown, black, and magenta modules) were highly stable with DEGs in F. graminearum affected silk ( Figure3C 1-2). It can also be seen that the average (“logFC value of control a + control b/2) log2 fold change value of differently expressed genes was used to construct the complex heat map ( Figure 3D ). The complicated heatmap representation of 6,668 core genes, logFC (DEGs), and GS (WGCNA) exhibited approximately equal Fg fungal infection ( Supplementary Table S4 ).
We extracted the protein–protein interaction network with a medium confidence score >0.4 of F. graminearum-affected Z. mays silk 6,668 core genes that matched with 3,325 STRING genes ( Supplementary Table S5 ). Network visualization and analysis were performed using Cytoscape, which identified 2,879 genes as nodes and 29,918 edges. No interaction was observed for the remaining 446 genes. Out of 86 clusters, only 12 MCODE scores were more significant than four of the MCODE clustering ( Tables 2 , S6 ). The PPI discovered 65 essential genes to be common in all four scoring techniques, such as MCC, MNC, EPC, and node–connect degree of CytoHubba ( Figure 4A ). Cluster 1 (35 nodes and 1,096 edges) contained downregulated genes, had the highest MCODE score (30.444), and overlapped the CytoHubba scoring. The enrichment analysis of the different biological processes of these essential genes was evaluated using the online enrichment tool ShinyGO. Substantial genes were enriched in photosynthesis, generation of precursor metabolites and energy, light reaction, and response to light stimulus ( Figure 4B ), indicating that a maximum number of genes were involved in the photosynthesis process.
A total of 14,601 DEGs were functionally identified in several fungus-affected samples. Overall, 773 high functional categories with 46.6, 45.7, 21.5, and 8.8% genes conforming to biological processes, molecular functions, cellular components, and pathways were identified, respectively ( Table 3 ). In F. graminearum (Fg) infections to the Z. mays silk, the up-regulated genes gene ontology (GO) enrichment analysis revealed biological process categories with the following GO terms: protein phosphorylation (347 genes of the DEGs), defense response (133 genes of the DEGs), phosphorylation (423 genes of the DEGs), phosphate-containing compound metabolic process (543 genes of the DEGs), cell surface receptor signaling pathway (50 genes of the DEGs), phosphorus metabolic process (544 genes of the DEGs), and response to biotic stimulus (83 genes of the DEGs), response to external biotic stimulus (75 genes of the DEGs). Upregulated genes likewise contain 179 biological processes, 149 molecular functions, eight cellular components, and nine functional pathway categories ( Supplementary Table S7 .FgU). The downregulated genes GO enrichment analysis revealed biological process categories with the following GO terms: polysaccharide metabolic process (124 genes of the DEGs), carbohydrate metabolic process (261 genes of the DEGs), cellular glucan metabolic process (79 genes of the DEGs), glucan metabolic process (80 genes of the DEGs), cellular polysaccharide metabolic process (91 genes of the DEGs), polysaccharide biosynthetic process (72 genes of the DEGs), photosynthesis, light harvesting in photosystem I (16 genes of the DEGs), cellular carbohydrate metabolic process (108 genes of the DEGs), cell wall organization or biogenesis (104 genes of the DEGs), photosynthesis, light reaction (40 genes of the DEGs), likewise, in down-regulated genes contain 149 of biological processes, 49 of molecular functions, 53 of cellular components, and 15 pathway functional categories ( Supplementary Table S7 .FgD). In Um infections to the Z. mays silk, GO enrichment analysis of the DEGs identified biological process categories with the following GO terms: regulation of RNA biosynthetic process (242 genes of the DEGs), regulation of RNA metabolic process (245 genes of the DEGs), regulation of nucleobase-containing compound metabolic process (247 genes of the DEGs), defense response (47 genes of the DEGs), regulation of cellular macromolecule biosynthetic process, regulation of macromolecule biosynthetic process (250 genes of the DEGs), regulation of cellular biosynthetic process, regulation of biosynthetic process (251 genes of the DEGs), protein phosphorylation (181 genes of the DEGs), and nucleic acid-templated transcription (246 genes of the DEGs). In the upregulated genes,110 genes were involved in biological process, 103 genes performed molecular functions, eight genes were responsible for cellular component, and four genes are responsible for pathways ( Supplementary Table S7 .UmU), and in downregulated genes, 95 were detected for biological process, 38 genes for molecular functions, 56 and 19 for cellular components, and 19 pathways, respectively ( Supplementary Table S7 .UmD). In Fv infections of the Z. mays silk, GO enrichment analysis of the DEGs identified biological process categories with the following GO terms: defense response (28 genes of the DEGs), cell surface receptor signaling pathway (13 genes of the DEGs), response to biotic stimulus (17 genes of the DEGs), response to bacterium (11 genes of the DEGs), defense response to other organisms (15 genes of the DEGs), protein phosphorylation (48 genes of the DEGs), response to oxidative stress (17 genes of the DEGs), response to external biotic stimulus, defense response to fungi (nine genes of the DEGs), cell wall polysaccharide catabolic process, xylan catabolic process (four genes of the DEGs), photosynthesis, light-harvesting (four genes of the DEGs), cell wall macromolecule catabolic process (four genes of the DEGs), and phenol-containing compound biosynthetic process (four genes of the DEGs). Upregulated genes contain 126 biological process, 93 molecular process, six cellular components, and five pathways ( Supplementary Table S7 .FvU), downregulated gene contain 31 biological processes, nine molecular functions, and 14 cellular components ( Supplementary Table S7 .FvD). GO terms were significantly enriched in Ta infections caused in Z. mays silk. The following GO terms were identified: response to the stimulus (six genes of the DEGs), response to stress (five genes of the DEGs), catabolic process (four genes of the DEGs), response to endogenous stimulus (three genes of the DEGs), regulation of metabolic process and regulation of cellular process (three genes of the DEGs), cellular response to stimulus (three genes of the DEGs), cell wall organization or biogenesis (two genes of the DEGs), regulation of cellular process (18 genes of the DEGs), regulation of metabolic process (16 genes of the DEGs), developmental process (seven genes of the DEGs), response to stimulus (five genes of the DEGs), cellular component organization (three genes of the DEGs), cell cycle process (three genes of the DEGs), and cellular component organization or biogenesis (two genes of the DEGs). Upregulated genes likewise contain 32 biological process, 21 molecular functions, and 14 genes for cellular components as identified ( Supplementary Table S7 .TaU), and the downregulated genes contain 33 biological processes, 21 molecular functions, and 17 cellular components with high-level GO categories ( Supplementary Table S7 .TaD).
The functional enrichment analysis of core genes with a false discovery rate <0.05 showed 704 higher-level GO categories in which 3,082, 3,254, 1,360, and 582 genes were involved in biological processes, molecular functions, cellular components, and pathways ( Supplementary Table S8 ). Core genes with highly enriched GO terms positively included biological processes to stimulus–response, defense responses against fungus, and phosphorylation ( Figure 5A ). The GO annotations of downregulated genes showed that they belong to different biological processes ( Figure 5B ) in Fg infection of Z. mays silk.
Among the 45 transcription factor (TF) classes, WRKY, NAC, ethylene-responsive factor (ERF), MYB, C2H2, basic helix–loop–helix (bHLH), and GRAS TFs were highly differentially expressed in infected silk. The heat shock transcription factor, transcription activator-like effectors, RAV, M-type-MADS, and ZF-HD showed up- and downregulation ( Figure 6A ). Out of 520 TFs from core genes, 410 TFs matched with ShinyGo functional annotation of DEGs. Three TFs, namely, bHLH-0, DRE-binding protein3/ERF, and MYB-110, were upregulated in all the infected samples. AP2-EREBP-115, C2C2-Dof-26, and Homeobox-59 were upregulated in Fg, Ta, and Um infections. Five WRKY, four NAC, and MYB, three bHLH, two AP2-EREBP, G2-like, and one bZIP TF family were upregulated, and two TFs, namely, bHLH-161 and Homeobox-60/71, were downregulated in Fg, Fv, and Um infections. In total, 228 TF genes were expressed in Fg and Um infections, while 147 TF genes were expressed in a Fg infection ( Supplementary Table S9 ). Core genes contained 257 genes for carbohydrate-active enzymes (CAZYme), while dbCAN2’s diamond, hmmer, and hotpep databases equally shared 169 and 88 up- and downregulated genes. Specific expressions of 16 and 10 modules were found in up- and downregulated genes, respectively, while 24 modules were found in both ( Figure 6B ). These genes belong to different modules, namely, glycosyl transferase (GT), glycoside hydrolase (GH), carbohydrate-binding modules (CBM), auxillary activities (AA), and carbohydrate esterases (CE). A further comparison with DEGs from other fungi (Fv, Ta, and Um) was conducted. The 20 CAZyme-related genes were detected in three fungus-infected silk samples (Fg, Fv, and Um), 123 genes in Fg and Um infections, and two genes in Fg and Fv infections; 112 CAZyme-related genes were only expressed in Fg infections of Z. mays silk ( Supplementary Table S10 ). Upon screening of resistance (R) genes from significant core genes, 346 and 174 up- and downregulated R-genes ( Supplementary Table S11 ) were found. These 520 R-genes are divided into 13 classes and seven domains. The majority of kinase and transmembrane (TM) domains comprised of the KIN class ( Figure 6C ), and the other classes were receptor-like kinases (RLK), receptor-like protein (RLP), receptor-like proteins consisting of an LRR repeat (RLP), contains coiled-coil and kinase (CK), nucleotide-binding site (N), CC-NBS-LRR (CNL), NL (NBS-LRRs), etc. Compared with Fv-, Ta-, and Um-infected silk, the putative DUF26-domain receptor-like protein kinase family protein showed a positive expression in all fungal infections. In the Fg, Fv, and Um conditions, 48 R-genes were considerably expressed. In the Fg and Um infections, 222 R-genes were expressed, with three R-genes in Fg and Fv infections and 227 R-genes expressed in Fg condition. The receptor-like serine/threonine-protein kinase, putative leucine-rich repeat receptor-like protein kinase family protein, and protein kinase superfamily were all present in more significant amounts in Fg and Um than in Fv and Ta conditions. These expression patterns played a crucial role during signal transduction and other biological functions.
Fungi are the second major biotic factor that reduce crop yield after insects. Some of the major fungal diseases of maize are Gibberella ear rot, Fusarium ear rot, corn smut, brown spot, etc., which cause considerable yield losses up to 42% (Thompson and Raizada, 2018). In addition, fungi produce many mycotoxins, leading to poisoning and quality deterioration (Agostini et al., 2019). However, there is a lack of research information regarding the direct comparative studies with respect to multiple fungal infections (Fg, Fv, Ta, and Um) in maize silk and identification of abundant genes in these four fungal infections. The current study is based on computational approaches of the publicly available transcriptome data of Z. mays silk infected with multiple fungi focused on core genes identification by differential expression analysis followed by co-expression analysis through WGCNA. We identified 14,694 and 14,808 DEGs of control datasets of A and B ( Supplementary Table S2 ). In further simplification, 4,519 and 5,125 genes were determined by comparing the up- and downregulated genes within controls. The up- and downregulated genes were compared to genes in the modules of WGCNA with relevant specific traits, and core genes were identified as described in the method and represented in Figure 3C . Our comparative study found that, in Fg infection conditions, more genes were affected compared to other Fv, Ta, and Um fungal infections. Twenty-one (21) most prominent genes identified in this study were expressed in all four fungal infections of maize silk. Many significant genes were identified, which were common to conditions caused by three and two fungi. Moreover, 4,120, 108, 68, and 760 genes were uniquely expressed in Fg-, Fv-, Ta-, and Um-affected silk ( Figure 2 ).
Twenty-one (21) DEGs were identified in all four infections ( Figure 2 ) which showed different expression values and functions in maize silk. The analysis of four fungus-affected samples revealed that these genes showed a higher significance in Fg infection than in other fungal infections. The upregulated expression of 21 genes was found except for the downregulation of CYP 450 in Fg, Fv, and Um and BP3, CRINKLY 4, OSM 34, DUF 26, and benzoxazinone in Ta along with CRINKLY 4 in Fv infections ( Figure 7 ). According to Li and co-workers, the peroxidase (POD) enzyme controls the lengthening of germ tubes to shield maize kernels from fungal diseases (Li et al., 2018). It is a fact that POD genes were found to be upregulated in all samples of maize silks infected by fungi, with logFC values of 11.02 in Fg infections, 8.7 in Fv infections, and 5.59 and 6.7 in Ta and Um infections, respectively. These are suggestive that the POD genes that we detected in our study also have a similar function. Interestingly, Um had a higher expression of the DBP3 protein gene than the other fungal infections in maize silks. Reports suggest that multiple steps downstream of the ABA-independent route show its significance in the regulation of abiotic stress (Joshi et al., 2016). Conversely, our research indicates that DREB protein synthesis promotes a defensive activity against fungal infections, especially Um. The expression of benzoxazinone is highest in the Fg and Fv tissues, followed by Um, and lowest in Ta infections. The current study coincides with the impact of another report which suggests that it has an effect on pests and antifungal activity (Cantillo et al., 2017). A DEG analysis revealed that, in Ta, the cytochrome p450 (CPY450) gene is upregulated despite showing downregulation in other fungal infections. This gene plays a variety of roles in plant defense, including the biosynthesis and catabolism of phytohormones and other secondary compounds (Xu et al., 2015; Lambarey et al., 2020; Li and Wei, 2020; Pandian et al., 2020). In addition, the expression pattern of CRINKLY4, a kinase family protein, showed a variable expression in fungus-affected silk. These proteins influence the shape of the cell size and the epidermal development in maize leaf (Becraft et al., 1996). CAT1 expression increases during microbial infections and hinders plant growth, according to a previous study (Yang et al., 2014). Our study demonstrated that fungal pathogen assaults on maize silks activated CAT1-related genes. The results also suggest that it could change the metabolic activity during fungal invasion in maize silks (Vina-Vilaseca et al., 2011; Yang et al., 2014). It is reported that SKIP19 protein genes influence and respond to biotic and abiotic stress and play a key role in soybean pollen tube germination and salt and drought tolerance (Chen et al., 2008; Yang et al., 2008; Chang et al., 2009; Ren et al., 2020). There is variable expression in Fg and Um infections, but the strong expression in Fv and Ta infections confirms the SKIP19 gene’s role in Z. mays Fv and Ta defense. The osmotic-like protein (OSM34) in plants, animals, and fungi improves host defense and immune defense against biotic and abiotic stress (de Jesús-Pires et al., 2020). Our research found that OSM34 protein genes were upregulated in Fg, Fv, and Um infections but downregulated in Ta infections in accordance with the roles mentioned. DUF26, which is upregulated in all fungal infections under study, except Ta, belongs to the receptor-like protein kinase sub-family; its domain plays a crucial role in stress resistance and antifungal defense (Liu et al., 2021). Putative RING zinc finger domain superfamily proteins have ubiquitin–protein ligase activity and help plant growth and development in A. thaliana (Gao et al., 2015; Kim et al., 2019). Small auxin-up RNA is a member of the auxin-responsive gene family that is upregulated in all the fungal infections considered in the present inquiry with logFC 11.5 in Fg, 8.5 in Fv, 4.7 in Ta, and 9.8 in Um. Previous research using microarray data profiling identified these genes as highly expressed in the roots and leaves but less in seeds, which is essential in plant growth and development (Chen et al., 2014; Zhang et al., 2021). S-norcoclaurine synthase proteins exhibit a defense response and have a signaling receptor activity, and four miscellaneous RNA genes were identified.
Within the Fg, Fv, and Um fungal infection conditions, 479 expressed genes out of 502 are intersectionally connected. These genes were relatively high in defense responses, photosynthesis, detoxification, and secondary metabolic processes. In any case, the Fg conditions have a higher expression value than those observed with Fv and Um infections ( Supplementary Table S12 ). Seventeen genes are highly expressed in Fg, Ta, and Um samples; these genes are involved in the DNA-binding transcription factor activity and are upregulated in these Fg and Um samples but downregulated in Ta infections ( Supplementary Table S12 ). Several genes implicated in the light reaction of photosynthesis were discovered to be highly expressed during abiotic stress conditions (McNinch et al., 2020). This finding clearly indicates that the presence of fungal pathogen in the Z. mays silk may be a crucial factor controlling the photosynthesis processes, functions of the plant cell surface receptor signaling pathway, and hydrogen peroxide catabolic process. In our study, the terpene synthase genes TPS6 (Ensembl id: Zm00001eb412960) and TPS11 (Ensembl id: Zm00001eb412980) were significantly upregulated in Fg, Fv, and Um infections caused in maize silk ( Supplementary Table S12 ). A group of researchers (Huffaker et al., 2011) observed that terpene synthase (TPS6 and TPS11) proteins involve the plant pathogen’s defense. TPS6 and TPS11 are transcribed only in the leaves and roots of Z. mays. TPS6/TPS11 function in terms of resistance to Um infections and tumor formations (van der Linde et al., 2011) and have a role in the production of several antibiotics (Huffaker et al., 2011).
During Fg and Um infections, a total of 3,838 were intersectionally connected. Seventy-seven genes (77) were found to be commonly expressed in Fg and Fv infections, while 20, 21, and 13 genes were common between Fg and Ta, Fv and Um, and Fv and Ta infections, respectively ( Supplementary Table S13 ). These genes were highly expressed in response to stimulus and stress. As suggested by Thompson and Raizada (2018), the maize silk have a defense mechanism against fungal infections, with wounds being the most susceptible to damage caused by Fg. Apart from maize silk, studies conducted by Reid et al. (1992) and du Toit and Pataky (1999) also identified that Fg and Um almost take the same time period for a successful infection in ear heads. Yang et al. (2018) suggested that E3 ligase under drought tolerance of Z. mays plays a crucial role in enabling plants to effectively and efficiently cope with environmental stress. In our results, RING-type E3 ubiquitin transferase was positively expressed under Fg- and Um-infected silk of Z. mays ( Supplementary Table S13 ). Photosynthesis, chlorophyll a-b binding protein, and light reaction photosynthesis I and II reaction center genes were highly downregulated. HVA22-like protein was downregulated, wherein HVA22 specifically inhibits GA-induced PCD/vacuolation of aleurone cells in barley (Guo and Ho, 2008). Glucanendo-1,3-beta-glucosidases (β-1,3-glucanases) protein genes have negative regulation, and these proteins play a significant role against the fungal pathogen by degradation of the cell wall. Lozovaya et al. (1998) found a positive correlation between the Aspergillus flavus fungus-infected kernel of maize silk and β-1,3-glucanases. Gao et al. (2017) found that GDSL esterase/lipase participates in immunity through lipid homeostasis in rice. In Fg and Um infection conditions, 10 GDSL genes were downregulated, and three upregulated genes were found. Huo et al. (2020) reported that 10 genes strongly contribute to male fertility, such as immature tassels, meiotic tassels, and others.
Maize silk infected with Fg activated more genes and was involved in phosphorylation ( Supplementary Table S14 ). This suggests that phosphorylation may be one of the initial events in a putative signal transduction pathway leading to the post-translational modification of a protein that controls cell cycle, development, growth, and stress responses. The research group of Palmer et al. (1993) reported that blue light induces phosphorylation in Z. mays plant mediated by an enzyme which belongs to the Ser/Thr class of kinases (Luan, 2002). Furthermore, a large number of genes were found to be involved in carbohydrate metabolic process when affected with the Fg pathogen, with the carbohydrate metabolism genes being downregulated during the process. When maize silk was infected with the Um pathogen, most genes responded to stimulus, stress, oxidation–reduction (redox) reactions, and biological processes like cell cycle (de la Torre et al., 2020). Genes involved in the cell cycle process were downregulated. It has been found that the cell cycle regulation and appressorium morphologenesis are delicately linked. The given primary function of the appressorium is to aid in the invasion of the plant tissue and the subsequent proliferation inside the host (de la Torre et al., 2020). Significant expression patterns were not observed with Fv and Ta infections. In summary, the Fg infections cause more damaging effects compared to other fungal infections. The analysis revealed that 4,355 were interconnected during intersectional studies ( Figure 2 ) with Fg and Um pathogen conditions belonging to Nectriaceae (Fg) and Ustilaginaceae (Um). During biotic stress, it was observed that genes associated with photosynthesis were downregulated as reported by researchers (Doke et al., 1996; Zhu and Li, 2015), which was in agreement with our results. In Fg infections, variations were observed with cell wall-related genes ( Supplementary Table S14 ), as fungal pathogens are known to secrete pectinases, xylanases, cellulases, and ligninases (Sharma, 2016) which can cause plant cell wall degradation during the infection process.
We identified 3,325 proteins from the string databases of core genes. The network was simplified into 12 highly sub-connected clusters ( Table 2 ) and identified essential proteins in the network based on the CytoHubba scoring method. The PPI network revealed that cluster 1 has 35 downregulated core proteins that infected the maize silk plant and were involved in biological processes ( Figure 4B ), such as photosynthesis. A research team (Horst et al., 2008) proposed that Um infection to Z. mays leaves reduced the photosynthetic rate and maintained the nutrients as well as influenced the chlorophyll content on a time scale (Kshirsagar et al., 2001). Wang et al. (2021) proposed the light harvesting in photosystem 1, a biosynthetic/metabolic process positively expressed in Gibberella stalk rot disease in Z. mays plant. Moreover, in other clusters 2 and 3, UMP pyrophosphorylase protein is involved in UMP biosynthesis via salvage and L-tryptophan biosynthesis. It has an intermediate role in benzoxazinoid biosynthesis with indole-3-glycerol phosphate in the chloroplast (Richter et al., 2021). Photorespiration, the pathway used to regenerate 2-phosphoglycolate metabolism, plays an essential role in photosynthesis in higher plants and is localized in chloroplasts (Eisenhut et al., 2008; Bräutigam and Gowik, 2016). Trehalose-6-phosphate synthase protein has a role in sugar-induced signaling pathway, and its function has different stages in the plant on growth and development. Iordachescu and Imai (2008) found that plant trehalose levels are typically low. They can change in response to shoot drought, salt, and cold stress challenges in roots and shoots. Another group of researchers (Henry et al., 2015) also reported that trehalose pathway genes were highly affected under saline conditions. Furthermore, these genes were downregulated with the involvement of fructose-bisphosphate aldolase (FBA) protein in various pathways, namely, glycolysis, carbohydrate degradation, and other physiological and biochemical processes. These biological processes include plant defense, response to biotic stress, plant growth, plant development, regulation of secondary metabolites, signal transduction, and Calvin cycle (Lv et al., 2017) and have been documented in other plant species including Z. Mays, A. thaliana, and Oryza sativa (Mininno et al., 2012) under abiotic stress conditions like salt, drought, heat and cold conditions. Our study finds FBA protein in cluster 2 and is downregulated in Fg infections in maize silk but upregulated in wheat to improve the enzyme activity and CO2 concentrations in green plant tissues during development (Lv et al., 2017). In cluster 4, upregulated genes involved phosphotransferase, which has a vital role in the hexose metabolism pathway, which is part of carbohydrate metabolism, to generate glucose-6-phosphate for glycolysis. GRMZM2G076075_P02, glucose-6-phosphate isomerase, is also involved in the glyconeogenesis process, whereas GRMZM2G161245_P01; malate dehydrogenase, is an enzyme that participates in the citric acid cycle from the conversion of malate into oxaloacetate (using NAD+) and also has a vice versa reaction (Takahashi-Íñiguez et al., 2016). In Pisum sativum, a 280% increase in malate dehydrogenase enzyme activity was observed with respect to Fusarium wilt diseases in comparison to control pea plants (Reddy and Stahmann, 1975) and downregulated proteins GRMZM2G074158_P01 and GRMZM2G085577_P01; α-1,4-glucan phosphorylase belongs to the glucosyltransferase family, and these enzymes have an important role in starch and metabolism pathway given the reversible transfer of glucosyl units from glucose-1-phosphate to the non-reducing end of α-1,4-d-glucan chains with the release of phosphate (Rathore et al., 2009). Cluster 6 has 45 nodes with 295 edges and one hub node, which is a cover scoring method. The expression of the protein GRMZM2G137151_P01 (1-deoxy-D-xylulose 5-phosphate synthase, DXS) genes was mainly in Artemisia annua leaf and flowering buds (Zhang et al., 2018a). In our study, the expression of this protein is upregulated in Fg fungal infections in the maize silks. A similar result was reported by Cordoba et al. (2011) who demonstrated plastid localization in Z. mays leaves. DXS catalyzes the first reaction that converts pyruvate and glyceraldehyde-3-phosphate to 1-deoxy-D-xylulose 5-phosphate in the methylerythritol phosphate pathway (Tambasco-Studart et al., 2005; Zhang et al., 2020). The remaining cluster 5 and nodes, namely, 7, 10, 11, and12 ( Table 2 ), do not cover up the pathways under the high scoring method within the top 100 ( Supplementary Table S15 ).
In our current endeavors, while analyzing the transcriptome data from different fungal infections and infections caused in corn silks, several TFs, R-gene, and CAZymes have been identified from core genes that serve as a molecular switch to interact with cis-acting transcription factor binding sites and directly control the transcriptional regulation of plant genes (Kimotho et al., 2019; Soni et al., 2020). In our study, 45 families were identified from 367 and 153 up- and downregulated transcripts of plant genes respectively ( Figure 6A ), such as WRKY, NAC, AP2/ERF, MYB, C2H2, bHLH, bZIP, etc. ( Supplementary Table S9 ). One of the significant plant-specific transcription factors is encoded by the WRKY gene family, discovered in several plant species (Ma et al., 2021) which have highly expressed genes. A team of researchers from the Louisiana State University (Fountain et al., 2015) reported similar results in maize with respect to the resistance and susceptibility to A. flavus fungal infection. WRKY plays a vital role in biotic stress and is involved in PAMP signaling and multiple defense responses through mitogen-activated protein kinase (MAPK) signaling, especially in sensing pathogen effectors or PAMP, and also interacts with resistance (R) protein (Meng and Zhang, 2013). Numerous studies have shown that a significant proportion of WRKY TFs are involved in disease response via the jasmonic acid (JA) signaling pathway (Ma et al., 2021). These TFs act as repressors or activators of basal defense responses (Windram et al., 2012). Similarly, NAC TFs participate in gene transcription regulations (Journot-Catalino et al., 2006), development, and stress response (Farhadian et al., 2021) and are the second highly expressed TFs in endosperm and kernels than in roots and stems that were known to regulate starch synthesis (Olsen et al., 2005; Xiao et al., 2021). Our study observed that NAC TFs were highly expressed in Fg infection of silk of Z. mays compared to other fungal infections such as Fv, Ta, and Um. Many plants have seen ERF TFs involved in disease resistance with phytohormone-mediated fungal defense (Luo et al., 2019), such as ERF activity in JA-mediated defense responses (Grennan, 2008; Jin et al., 2017). In A. thaliana, DREB TFs represent a large part of the AP2/ERF superfamily (Agarwal et al., 2017; Hrmova and Hussain, 2021). An MYB transcription factor is a more prominent family involved in multiple biological functions in the plants, such as primary and secondary metabolite reactions, regulating the plant growth and development, cell morphogenesis, and response to biotic and abiotic stress (Cao et al., 2020; Duan et al., 2021). However, in plants, MYB TF works as an activator for transcription that triggers G2/M-specific gene expression (2011; Haga et al., 2007). Our findings indicated that MYB-related genes might be involved in the Z. mays pathogen response since the expression profile of the MYB-related gene family in Z. mays and soybeans exhibits a wide range of variation with time following a Um infection (Du et al., 2013). Researchers studied the expression of bHLH TFs in the young leaf, root, and auricular tissue of Z. mays which have a high expression while being involved in plant development (Murre et al., 1994; Zhang et al., 2018b). After about a decade, the bHLH TF was identified in a study conducted on A. thaliana that acts as the target of JAZ protein and negatively regulates JA-mediated plant defense and development (Song et al., 2013). Wei et al. (2012) reported that 125 bZIP genes were found, which encode 170 proteins in Z. mays tissue; 18 bZIPTF were significantly upregulated as expressed in silk. These TFs regulate different biological processes such as floral development, seed formation, response to biotic and abiotic stress (Katagiri et al., 1989), starch synthesis in rice endosperm and maize kernel, and they saw that starch synthesis genes have a similar expression pattern (Wang et al., 2013). The invasion of fungal pathogens through penetration to the silk of Z. mays enhanced the defense mechanism against pathogens through activation of the gene encoding cell wall-associated proteins, of which UGTs (UDP-glycosyltransferases) have been found in Z. mays and other species in investigations (Duan et al., 2021). It is involved in the production of phytohormones, metabolites, growth, development, and biotic and abiotic stress (Li et al., 2014; Rehman et al., 2018). UDP members of the GT1 family catalyze, help the biosynthesis of oligo- and polysaccharides, and transfer sugar residues from nucleotide donor substrates to receptor substrates or a developing carbohydrate chain (Hoffmeister et al., 2001; Jayaprakash et al., 2021). GT1-related genes were substantially up-regulated in the Fg-affected silk of Z. mays. Cao et al. (2008) have reported that the GT1 family is the most prominent family in all three species. Aside from the GT1 family, rice’s top five GT families include the GT2, GT4, GT8, GT31, and GT47 families in A. thaliana and Populus species (poplar). GT8 and GT47 classes were highly downregulated in Fg fungal infections (Kong et al., 2019), and GT8 family was found to play an essential role in plant cell wall formation which is considered critical for growth and development. The cold and saline conditions significantly caused upregulation to aid in salt stress tolerance. Cao et al. (2008) distinguish that most GT47 genes have a low expression in the different development stages of rice. Similarly, the second most highly expressed CAZyme family was GH ( Supplementary Table S10 ), which is involved in the carbohydrate metabolic process (GO:0005975) to cleave glycosidic bonds in various forms of glucan, glycosides, and glycoconjugates. These also have industrial use and biotechnological applications to develop bio-fuel (xylanases, cellulases, etc.) and are useful in pharmaceutical research (Roy et al., 2020). This study also identified CAZymes like CBM, CEs, and AA enzyme to be involved in lignin catabolic process with less gene expression in differential expression analysis ( Figure 6B ). We were also successful in identifying 13 classes of R-genes which consist of seven different domains from core genes. In this particular analysis, the KIN class was found to be a major class of R-genes consisting of kinase (KIN) and kinase transmembrane KIN-TM domains ( Figure 6C ). Similarly other major R-gene classes identified were receptor-like kinases (RLK), receptor-like protein (RLP), and receptor-like-protein consisting of a leucine-rich repeat (LRR), which play a crucial role in plant development as well as response to biotic and abiotic stress (De Hoff et al., 2009). Resistance genes are also known as adult plant resistance genes or quantitative resistance genes. R-genes have been identified within host and pathogen cells (Jones et al., 2014). The defense response in plants depends on the pathogen’s attack, such as phytohormones involved in defense responses and salicylic acid which controls the biotrophic pathogens (Kelley et al., 2012). Jasmonic acid (JA)-dependent and ethylene (ET)-dependent signaling pathways regulate the necrotrophic pathogens (Bari and Jones, 2009; Birkenbihl and Somssich, 2011). Most plant disease R proteins have been seen to contain a series of LRRs, a nucleotide-binding site (NBS), and a putative amino-terminal signaling domain. Thus, NBS–LRR protein activations are a phenomenon that changes the structure as well as nucleotide-binding status (DeYoung and Innes, 2006). DUF26 is a subfamily receptor-like protein kinase. It has 90% identity with the cysteine-rich receptor (CRR)-like protein kinase; its domain has antifungal activity and an essential role in stress resistance (Liu et al., 2021). It has been discovered in Arabidopsis plant that the CRR RLKs (CRKs) include two DUF26, CRK9, CRK26, and four DUF6 domains. These domains are involved in ABA signaling via regulating the ABA responses to seed germination, development, abiotic stress, and potential antifungal agents (Quezada et al., 2019). DUF26 domain protein genes were positively expressed in our study in all infected fungal conditions. Two proteins of DUF26 domain—AFP1 and AFP2—were found in Um-infected Z. mays apoplastic fluid (Ma et al., 2018) which are upregulated and act as an antifungal, thus increasing the resistance to fungal pathogens. In total, 48 R-genes are significantly expressed in Fg, Fv, and Um conditions, 222 R-genes have been expressed in Fg and Um conditions, three R-genes were found in Fg and Fv infections, and 227 R-genes are described in Fg condition. Subsequently, R-genes such as protein kinase superfamily proteins, receptor-like serine/threonine-protein kinase (EC 2.7.11.1), and putative leucine-rich repeat receptor-like protein kinase family protein were more significantly expressed in Fg and Um compared to other fungal infections (Fv and Ta). Alam et al. (2010) reported that LRR–RLK gene expression was significantly low in a salt-stress condition compared to other abiotic stresses. Additionally, putative receptor-like protein kinase, leucine-rich repeat protein kinase family protein, disease resistance RPP13-like protein 4, L-type lectin-domain containing receptor kinase IX.1, and LRR family proteins were significantly expressed only in Fg and Um ( Supplementary Table S11 ). It is known that the LRR sequence participates in a strong PPI.
A crucial clue to biotic variables affecting the Z. mays plant was discovered after the available silk of Z. mays transcriptome information was unified on NCBI and re-examined. Based on a comparative study using statistical estimates such as the DEG, the expression level of genes and the behavior of similar coding proteins have been found very distinct during multiple fungal infections in the silk of Z. mays. The 21 DEGs have been found in all four fungus-infected silk of Z. mays. The up- and downregulated genes were compared to the genes in the modules of WGCNA with relevant specific traits and identified core genes. In the current study, 520 from TFs, 169 from CAZyme, and 520 from R-genes among 6,668 core genes are involved in Fg infection in Z. mays silk, but not in the other fungal systems examined (Fv, Ta, and Um) for their transcriptomic datasets of Z. mays silk. The present study supports that the immunity stimulus and resistance genes are upregulated, while the downregulated genes are involved in photosynthesis, cell wall organization, pectin metabolism process, and response to auxin in the silk of Z. mays. We know that Fg and Fv belong to Nectriaceae family, but gene expression is very different, with 723 in Fv and 8577 in Fg. It is reported that Fg has an antagonistic relationship with Um, but the current study supports the probable similar function of the expressed genes during a fungal infection. Based on the transcriptome data analysis, we found that Ta does not affect the infected maize silk.
Publicly available datasets were analyzed in this study. The names of the repository/repositories and accession number(s) can be found in the article/supplementary material.
AK, KRK, and PTVL designed the research. AK and KRK performed the analysis and analysed the results. AK wrote the initial manuscript. PTVL, RSD, and AA majorly reviewed and analysed the draft and finalized the manuscript. All authors contributed to the article and approved the submitted version.
The fund for this study was received from Pondicherry University.
The authors are thankful to the Department of Bioinformatics, Pondicherry University for providing all the necessary infrastructure for this work. AK is grateful to Pondicherry University for providing a Non-National Eligibility Test fellowship.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647131 | Haigang Ma,Yongjiang Liu,Xueyan Zhao,Suhong Zhang,Hongxiang Ma | Exploring and applying genes to enhance the resistance to Fusarium head blight in wheat | 27-10-2022 | wheat disease,Fusarium graminearum,Fusarium head blight,genetics,breeding | Fusarium head blight (FHB) is a destructive disease in wheat worldwide. Fusarium graminearum species complex (FGSC) is the main causal pathogen causing severe damage to wheat with reduction in both grain yield and quality. Additionally, mycotoxins produced by the FHB pathogens are hazardous to the health of human and livestock. Large numbers of genes conferring FHB resistance to date have been characterized from wheat and its relatives, and some of them have been widely used in breeding and significantly improved the resistance to FHB in wheat. However, the disease spreads rapidly and has been severe due to the climate and cropping system changes in the last decade. It is an urgent necessity to explore and apply more genes related to FHB resistant for wheat breeding. In this review, we summarized the genes with FHB resistance and mycotoxin detoxication identified from common wheat and its relatives by using forward- and reverse-genetic approaches, and introduced the effects of such genes and the genes with FHB resistant from other plant species, and host-induced gene silencing (HIGS) in enhancing the resistance to FHB in wheat. We also outlined the molecular rationale of the resistance and the application of the cloned genes for FHB control. Finally, we discussed the future challenges and opportunities in this field. | Exploring and applying genes to enhance the resistance to Fusarium head blight in wheat
Fusarium head blight (FHB) is a destructive disease in wheat worldwide. Fusarium graminearum species complex (FGSC) is the main causal pathogen causing severe damage to wheat with reduction in both grain yield and quality. Additionally, mycotoxins produced by the FHB pathogens are hazardous to the health of human and livestock. Large numbers of genes conferring FHB resistance to date have been characterized from wheat and its relatives, and some of them have been widely used in breeding and significantly improved the resistance to FHB in wheat. However, the disease spreads rapidly and has been severe due to the climate and cropping system changes in the last decade. It is an urgent necessity to explore and apply more genes related to FHB resistant for wheat breeding. In this review, we summarized the genes with FHB resistance and mycotoxin detoxication identified from common wheat and its relatives by using forward- and reverse-genetic approaches, and introduced the effects of such genes and the genes with FHB resistant from other plant species, and host-induced gene silencing (HIGS) in enhancing the resistance to FHB in wheat. We also outlined the molecular rationale of the resistance and the application of the cloned genes for FHB control. Finally, we discussed the future challenges and opportunities in this field.
Fusarium head blight (FHB), which is also known as head scab and ear blight, caused by Fusarium graminearum (teleomorph Gibberella zeae) species complex is a fungal disease responsible for severe yield losses and poor grain quality in wheat (Triticum aestivum L.) (Bai and Shaner, 2004; Xu and Nicholson, 2009). The pathogen also produces mycotoxins such as trichothecenes and zearalenone contaminating infected wheat grains, which are harmful to humans and animals (Chen et al., 2019). Due to the warm temperature, abundant rainfall, and maize/wheat and rice/wheat rotations, the disease has been frequent and severe for the last decade worldwide, especially in China (Bai et al., 2018; Ma et al., 2019). The most effective and economical solution for reducing FHB damage is to identify genes related to FHB resistance and apply them to breed disease-resistant varieties. The resistance to FHB is quantitative in wheat and no immune genes have been found so far (Bai et al., 2018). To date, a substantial number of quantitative trait loci (QTL) or genes conferring FHB resistance have been reported (Liu et al., 2009; Zheng et al., 2021). Previously, strategies and progress of wheat breeding for FHB resistance have been reviewed (Ma et al., 2019; Zhu et al., 2019). Here, we summarize advances in wheat resistance to FHB with the main focus on the characterized genes related to FHB resistance and their function in genetic improvement for the FHB resistance in wheat.
Of the hundreds of QTL identified for FHB resistance by molecular mapping in common wheat, Fhb1, a QTL located on the short arm of chromosome 3B with the largest explanation of phenotype variation, provides durable and stable resistance to FHB. Fhb1 candidate genes have been cloned recently using the map-based cloning approach. In 2016, a pore-forming toxin-like (PFT) gene was firstly cloned as the candidate of Fhb1 (Rawat et al., 2016). However, this gene was also found in some susceptible accessions without Fhb1 (Yang et al., 2005; He et al., 2018a; Jia et al., 2018). Before long, another gene named HRC or His was cloned as an Fhb1 candidate by two independent studies (Li et al., 2019; Su et al., 2019). HRC/His encoded histidine-rich calcium-binding protein located in the nucleus. In comparison to that in the susceptible lines (HRC/His-S), the gene in the resistant lines carrying Fhb1 (HRC/His-R) had a deletion in its genome, which is responsible for FHB resistance (Li et al., 2019; Su et al., 2019). The function of TaHRC was validated by using a BSMV-mediated gene editing system in Bobwhite and Everest (Chen et al., 2022a; Chen et al., 2022b). It was recently found that HRC/His-S from Leymus chinensis (named LcHRC in the original article), which showed identical amino acid sequence to wheat HRC/His-S, bound calcium and zinc ion in vitro (Yang et al., 2020). Arabidopsis thaliana seedlings overexpressing LcHRC showed sensitivity to abscisic acid (ABA) (Yang et al., 2020). A protein that participates in heterochromatin silencing was identified as an LcHRC interactor through the screening of Arabidopsis yeast cDNA library (Yang et al., 2020). These results suggest a potential role of LcHRC in the regulation of genes involved in abiotic stress response. In wheat, HRC-S interacting proteins were identified through the screening of wheat yeast cDNA library (Chen et al., 2022b). One of the interactors, TaCAXIP4 [a cation exchanger (CAX)-interacting protein 4], was further validated to physically interact with HRC-S in planta (Chen et al., 2022b). The interaction with HRC-S suppressed TaCAXIP4-mediated calcium cation (Ca2+) transporting in yeast cells and resulted in reduced reactive oxygen species (ROS) triggered by chitin (Chen et al., 2022b), leading to the hypothesis that Ca2+ signaling-mediated ROS burst is essential for wheat FHB resistance. However, the details on how HRC/His affects FHB resistance remain equivocal, and more efforts are needed to elucidate their biological function and the regulatory network they mediated in defense response. In fact, Fhb1 locus has been widely used for FHB resistance breeding prior to the gene cloning. A large number of wheat varieties worldwide carried Fhb1 locus, which confers moderate FHB resistance with the reduction of at most 50% in FHB severity (Bai et al., 2018; Zhu et al., 2019), which further confirmed the solid role of this locus in FHB resistance.
Besides Fhb1, many other QTL conferring FHB resistance of wheat have been reported, but the majority of their candidate genes remain unidentified. QFhb.mgb-2A, a major QTL located on chromosome 2A, was found in a recombinant inbred line (RIL) population, obtained by crossing an hexaploid line derived from a resistant cultivar Sumai3 and a susceptible durum cv. Saragolla (Giancaspro et al., 2016). Several genes, including Fatty Acyl-CoA Reductase 1, Wall-associated receptor kinase 2 (WAK2), Arginine decarboxylase, SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily A member 3, and Ubiquitin thioesterase otubain genes, were detected in the QTL region (Gadaleta et al., 2019). The homeolog of WAK2 in common wheat, which was named TaWAK2A-800, was identified later as a positive regulator of wheat resistance to FHB (Guo et al., 2021). Knocking down TaWAK2A-800 in wheat using the virus-induced gene silencing (VIGS) method compromised FHB resistance, which may be attributed to the impaired defense pathway induced by chitin (Guo et al., 2021).
Decades of efforts in plant immunity have led to the development of plant resistance gene pool and the understanding of the mechanisms of plant disease resistance. The completion of wheat genome sequencing provides great convenience for the identification of the homologs of resistance genes in wheat through genome-wide homologous sequence analysis. Moreover, a variety of omics methods including transcriptomics, proteomics, and metabolomics will aid in identifying wheat resistance genes. A number of reverse genetics techniques are applied subsequently to overexpress and/or knock out/knock down the identified genes with the aim to verify their function. The widely used approaches in wheat include clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9)-based genome editing, VIGS, and RNA interference (RNAi). They have been invaluable in analyzing gene function in wheat FHB resistance.
Many classes of genes have been implicated in the resistance to FHB in wheat. One group of them is referred to as pathogenesis-related (PR) genes. Increased expression of PR genes is a hallmark of plant defense response to pathogen attack. Based on gene sequence homology, two wheat PR genes, encoding chitinase (PR3) and β‐1,3‐glucanase (PR2), respectively, have been separately overexpressed in wheat and enhanced FHB resistance was observed in greenhouse but not in the field when using inoculated corn kernels (Anand et al., 2003). In another study, transgenic wheat lines with overexpression of wheat α-1-purothionin gene (a PR gene) exhibited increased FHB resistance in the field condition when using the spaying inoculation method (Mackintosh et al., 2007). The essential role of plant hormones in the disease resistance is also a global consensus. Several genes involved in wheat phytohormone biosynthesis or signaling have been identified for FHB resistance. EIN2 is a central regulator of ethylene (ET) signaling (Alonso et al., 1999). RNA interference (RNAi)-mediated EIN2 silencing in wheat (Travella et al., 2006) reduced FHB symptoms (Chen et al., 2009), implying that ET signaling may promote wheat susceptibility to F. graminearum. Likewise, auxin was also implicated in FHB susceptibility of wheat. The expression of an auxin receptor gene TaTIR1 was found to be downregulated during F. graminearum infection (Su et al., 2021). Knockdown of TaTIR1 in wheat using RNAi technology increased FHB resistance (Su et al., 2021). In addition to phytohormones, wheat metabolites are also essential for FHB resistance. Using a metabolomics approach, a research group identified several genes conferring resistance to FHB, including TaACT encoding agmatine coumaroyl transferase (Kage et al., 2017a), TaLAC4 encoding laccase (Soni et al., 2020), and TaWRKY70 and TaNAC032 both encoding transcription factors (Kage et al., 2017b; Soni et al., 2021). These genes are all involved in the biosynthesis of hydroxycinnamic acid amides and phosphotidic acid, the major metabolites accumulated in wheat rachis after F. graminearum invasion. Suppressing the expression of these genes respectively using VIGS reduced FHB resistance. Transcriptomics are powerful in identifying genes related to FHB resistance. Lots of genes whose expression are induced by F. graminearum have been identified by different methods, such as Genechips and RNA sequencing, and some of them have been validated to be effective in FHB control. A wheat orphan gene named T. aestivum Fusarium Resistance Orphan Gene (TaFROG), which is a taxonomically restricted gene specific to the grass subfamily Pooideae, was identified as an F. graminearum-responsive gene and promoted wheat resistance to FHB (Perochon et al., 2015). TaFROG encodes a protein with unknown function but binds to TaSnRK1α, which is a wheat α subunit of the Sucrose Non-Fermenting1 (SNF1)-Related Kinase1 and plays central roles in plant energy and stress signaling (Perochon et al., 2015). Another TaFROG interactor, T. aestivum NAC-like D1 (TaNACL-D1), which is a NAC [No apical meristem (NAM), Arabidopsis transcription activation factor (ATAF), Cup-shaped cotyledon (CUC)] transcription factor, was identified by using yeast two-hybrid screening (Perochon et al., 2019). TaNACL-D1 was also responsive to F. graminearum and enhanced wheat resistance to FHB with unclarified mechanisms (Perochon et al., 2019). Plant lectins, a class of proteins binding reversibly to mono- or oligosaccharides, are often associated with biotic and abiotic responses. Wheat genes encoding lectins have been shown to improve FHB resistance. TaJRLL1 and Ta-JA1/TaJRL53 are two genes in wheat encoding jacalin-related lectins. Suppressing their expression in wheat separately using VIGS compromised the disease resistance to FHB, while overexpressing TaJRL53 in wheat enhanced FHB resistance (Ma et al., 2010a; Xiang et al., 2011; Chen et al., 2021). Other genes, including TaLRRK-6D encoding a leucine-rich repeat receptor-like kinase (Thapa et al., 2018), TaMPT encoding a mitochondrial phosphate transporter responsible for transporting inorganic phosphate (Pi) into the mitochondrial matrix (Malla et al., 2021), TaSAM encoding an S-adenosyl methionine (SAM)-dependent methyltransferase that catalyzes the transfer of methyl groups from SAM to a large variety of acceptor substrates (Malla et al., 2021), TaPIEP1 encoding transcription factor (Liu et al., 2011), and TaSHMT3A-1 encoding serine hydroxymethyltransferase (Hu et al., 2022), are also identified to contribute to FHB resistance. The abovementioned genes with FHB resistance are listed in Table 1 . They varied enormously in gene products and biochemical functions. It seems that wheat utilizes extensive biological processes to defend against F. graminearum attack. As no immune genes were found in FHB resistance, a deep understanding of the signaling pathway mediated by these resistance genes will help to optimize wheat FHB resistance in breeding.
Plant pathogenic microbes always secrete proteins that act as effectors into host cells to evade or inhibit host immunity, leading to enhanced pathogen virulence and facilitated pathogen growth (Dou and Zhou, 2012). The secreted effectors bind host proteins to modify their native biological functions. Some of the host targets are key regulators of plant immunity and therefore could be deployed for disease control. Secreted proteome of F. graminearum has been obtained, with the protein numbers varied in different studies (Yang et al., 2012; Rampitsch et al., 2013; Yang et al., 2013; Lowe et al., 2015). However, their host targets are largely unknown. It has been found that F. graminearum produces orphan secretory proteins (OSPs), and one of them, Osp24, functions as an effector (Jiang et al., 2020). After being secreted into wheat cells, Osp24 binds wheat protein TaSnRK1α (Jiang et al., 2020). The binding by Osp24 accelerates TaSnRK1α degradation, which may suppress host defense responses including cell death and is thus beneficial for pathogen infection; however, physical interaction with TaFROG, a wheat orphan protein, prevents TaSnRK1α from degradation and helps in wheat defense (Jiang et al., 2020). The interplay between the two orphan genes, OSP24 and TaFROG, may be indicative of co-evolution of F. graminearum and the host wheat, and the distinctive defense response of wheat to F. graminearum.
The mycotoxins such as deoxynivalenol (DON, a type B trichothecene) produced by the pathogen are toxic to humans and animals. They cause emesis, feed refusal, and even death (Eriksen and Pettersson, 2004). In addition, DON is considered as a virulence factor capable to facilitate disease spread on wheat (Proctor et al., 1995; Bai et al., 2002). F. graminearum deficient in DON biosynthesis was able to infect wheat spikelets but failed to spread in spikelets, thus causing diminished disease symptoms (Bai et al., 2002). Therefore, decreasing the amount of DON of wheat grain during pathogen infection is not only necessary for food security, but also one goal of breeding for FHB resistance. Proteins encoded by various genes have been identified with the ability to detoxify DON ( Table 1 ). Among them, uridine diphosphate (UDP)-glycosyltransferases (UGTs) have been widely reported to be able to detoxify DON through glucosylation. These enzymes transfer a glycosyl group from UDP-glucose to DON to conjugate DON into deoxynivalenol-3-O-glucose (D3G), which is nontoxic for animals. As DON can promote disease spreading, glucosylation of DON to D3G is an important plant defense mechanism. He et al. (2018b) systematically analyzed family-1 UGTs and identified 179 putative UGT genes in a reference genome of wheat, Chinese Spring. Among them, TaUGT3 (Ma et al., 2010b; Pei, 2011; Chen, 2013), TaUGT5 (Zhao et al., 2018), and TaUGT6 (He et al., 2020) were validated to be effective in reducing DON content in wheat. Wheat lines overexpressing the three genes respectively showed resistance to DON treatment and the resultant disease resistance to FHB, implying the potential of TaUGT as useful disease resistance genes in breeding for FHB resistance. Adenosine triphosphate (ATP)-binding cassette (ABC) transporters have been implicated in DON detoxication. They may export DON from the cytoplasm to reduce the damage caused by mycotoxin. TaABCC3, encoding an ABC transporter responsible for substance transport across cell membrane, was cloned from DON-treated wheat transcripts (Walter et al., 2015). Inhibition of TaABCC3 expression by VIGS increased wheat sensitivity to DON (Walter et al., 2015). However, the effect of TaABCC3 on FHB resistance was not analyzed. TaPDR1 and TaPDR7, two wheat genes encoding the pleiotropic drug resistance (PDR) subfamily of ABC transporters, were upregulated by DON treatment and F. graminearum infection; knockdown of TaPDR7 in wheat by VIGS compromised FHB resistance (Shang et al., 2009; Wang et al., 2016). Cytochrome P450, membrane-bound enzymes that can perform several types of oxidation–reduction reactions, was also reported to possess the ability to catabolize DON (Ito et al., 2013). A wheat P450 gene, TaCYP72A, was found to be activated by DON treatment and F. graminearum infection (Gunupuru et al., 2018). Suppressing TaCYP72A through VIGS reduced wheat resistance to DON (Gunupuru et al., 2018). However, whether this gene confers FHB resistance is not identified.
There are over 300 species classified under more than 20 genera in Triticeae (Dewey, 1984), which represent an invaluable gene pool for wheat improvement. The wild relatives of wheat are an important source for wheat improvement with FHB resistance. Many genes with FHB resistance have been identified and verified in vivo ( Table 2 ). They show unique features as well as shared characteristics with those identified in hexaploidy wheat.
Of the QTL that showed a stable major effect on FHB resistance, Fhb7 was transferred from wheatgrass Thinopyrum (Fu et al., 2012; Guo et al., 2015), and was cloned recently using the map-based cloning approach (Wang et al., 2020a). Fhb7 was mapped to chromosome 7E of Th. elongatum (Fu et al., 2012; Guo et al., 2015). The underpinning gene of the locus was identified that encodes a glutathione S-transferase (GST) with the prominent ability to detoxify trichothecene toxins produced by the pathogens (Wang et al., 2020a). The expression of the gene was increased at the late stage of infection and was also induced by trichothecene treatment (Wang et al., 2020a), implying an active role of Fhb7 in the response to mycotoxins. How Fhb7 is regulated at the molecular level remains obscure and needs to be determined. This may help to increase the expression of Fhb7 in wheat cultivars for further enhanced FHB resistance. Notably, wheat lines with Fhb7 locus showed increased resistance to FHB without growth defect and yield penalty (Wang et al., 2020a), making Fhb7 a promising potential for wheat resistance breeding.
Barley (Hordeum vulgare) PR genes also contributed to FHB resistance. Transgenic wheat lines separately overexpressing barley tlp-1 and PR2 gene showed increased resistance to FHB (Mackintosh et al., 2007). Two other genes, HvWIN1 encoding a transcriptional regulator of cuticle biosynthetic genes and HvLRRK-6H encoding a leucine-rich receptor-like kinase, were identified as positive regulators of FHB resistance (Kumar et al., 2016; Thapa et al., 2018). Knockdown of the two genes individually by VIGS increased the disease severity of barley (Kumar et al., 2016; Thapa et al., 2018). Wheat lines overexpressing a barley chitinase gene improved wheat resistance to FHB (Shin et al., 2008). UGT genes responsible for DON detoxication have also been identified in barley. HvUGT13248 played an effective role in DON detoxication when expressed in yeast (Schweiger et al., 2010), Arabidopsis (Shin et al., 2012), durum (Mandalà et al., 2019), and wheat (Li et al., 2015; Li et al., 2017; Mandalà et al., 2019). Wheat lines constitutively expressing HvUGT13248 showed improved FHB resistance (Li et al., 2015; Li et al., 2017; Mandalà et al., 2019). Recombinant HvUGT-10W1 purified from bacterium cells inhibited hypha growth of F. graminearum in the PDA (potato/dextrose/agar) media. Furthermore, suppressing HvUGT-10W1 expression in a barley variety 10W1, which showed resistance to FHB, using the VIGS approach reduced the resistance to FHB (Xing et al., 2017), implying the positive role of HvUGT-10W1 in barley resistance to FHB.
CERK1-V (Dv07G125800), the chitin-recognition receptor of Haynaldia villosa, was recently cloned and introduced into wheat under the drive of maize ubiquitin promoter (Fan et al., 2022). The overexpression lines showed enhanced FHB resistance, implying that the perception of chitin is an important step to initiate FHB resistance. Therefore, it has potential to identify the genes involved in chitin signaling and develop them for FHB resistance.
Brachypodium distachyon has been developed for FHB resistance analysis (Peraldi et al., 2011). Several genes from B. distachyon have been characterized by FHB resistance. Bradi5g03300 UGT gene has been introduced into B. distachyon Bd21-3 and the wheat variety Apogee, both of which are susceptible to FHB (Schweiger et al., 2013; Pasquet et al., 2016; Gatti et al., 2019). Enhanced resistance to FHB and strong reduction of DON content in infected spikes were observed in the transgenic lines. Promisingly, some of the transgenic lines with high Bradi5g03300 transcripts showed normal growth or phenotype compared with the wild type. BdCYP711A29 (Bradi1g75310) encoding cytochrome P450 monooxygenase involved in orobanchol (one form of strigolactones) biosynthesis was identified to negatively regulate FHB resistance. Overexpression of BdCYP711A29 in B. distachyon increases susceptibility to FHB, while the TILLING mutants showed disease symptoms similar to those of the wild type (Changenet et al., 2021).
Arabidopsis has been exploited for the analysis of the scientific rationale of plant resistance to F. graminearum because the fungi can infect Arabidopsis flowers (Urban et al., 2002; Brewer and Hammond-Kosack, 2015). Many genes that have been identified from A. thaliana showed potential for resistance against these pathogenic fungi. NPR1 is an ankyrin repeat-containing protein involved in the regulation of systemic acquired resistance. Wheat lines overexpressing NPR1 showed enhanced resistance to FHB (Makandar et al., 2006). AtALA1 and AtALA7, two members of Arabidopsis P-type ATPases, contributes to plant resistance to DON through cellular detoxification of mycotoxins (Wang et al., 2021). They mediated the vesicle transport of toxins from the plasma membrane to vacuoles. Transgenic Arabidopsis or maize plants overexpressing AtALA1 enhanced resistance to DON and disease caused by F. graminearum. It remains unknown whether AtALA1 homologous genes exist in wheat genome and have the same detoxification function.
F. graminearum also infects other crops, such as rice (Oryza sativa L.) and maize (Zea mays L.). In rice plants, a UGT OsUGT79 expressed and purified from bacterium cells was reported to be effective in conjugating DON into D3G in vitro (Michlmayr et al., 2015; Wetterhorn et al., 2017) and could be used as a promising candidate for FHB resistance breeding. Additionally, overexpression of rice PR5 gene encoding thaumatin‐like protein in wheat reduced FHB symptoms (Chen et al., 1999). In maize, RIP gene b-32 encoding ribosome inactive protein promotes FHB resistance when overexpressed in wheat (Balconi et al., 2007).
Host-induced gene silencing (HIGS) was recently developed to control fungal diseases, in which transgenic host plants produce small interference RNAs (siRNAs) that match important genes of the invading pathogen to silence fungal genes during infection (Machado et al., 2018). Koch et al. (2013) reported that detached leaves of both transgenic Arabidopsis and barley plants expressing double-stranded RNA from cytochrome P450 lanosterol C-14a-demethylase genes exhibited resistance to F. graminearum. HIGS transgenic wheat targeting the chitin synthase 3b also confers resistance to FHB (Cheng et al., 2015). HIGS targeting multiple genes involving FgSGE1, FgSTE12, and FgPP1 of the fungus is effective and can be used as an alternative approach for developing FHB- and mycotoxin-resistant crops (Wang et al., 2020b). Spray-induced gene silencing (SIGS), mechanistically similar to HIGS, is also effective to fungal disease control. In this approach, sprayed siRNAs or noncoding double-stranded (ds)RNAs onto plant surfaces targeting key genes of pathogens are taken up by the pathogens and in turn inhibit pathogen gene expression, leading to inhibited pathogen growth in plants (Koch et al., 2016; Song et al., 2018). SIGS targeting F. graminearum genes has been reported to effectively inhibit the pathogen growth in barley, providing the potential for FHB control (Koch et al., 2019; Werner et al., 2020; Rank and Koch, 2021).
Compared with agronomic practices, chemical control, and biological control, genetic resistance is the best and most cost-effective strategy that could provide meaningful, consistent, and durable FHB control (Shude et al., 2020). There are variations in the susceptibility of different host plant species to FHB; however, no wheat varieties possess immunity against FHB (Dweba et al., 2017). Though hundreds of QTL have been reported in wheat, only two, Fhb1 and Fhb7, have been cloned through years of hard work (Rawat et al., 2016; Li et al., 2019; Su et al., 2019; Wang et al., 2020a); thus, the FHB resistance genes that can be used for breeding is obviously limited. How to improve FHB resistance to a high level in wheat using the limited genes is a fundamental ongoing challenge. Pyramiding resistance genes to increase FHB resistance is feasible and popularized. However, the strategy is highly dependent on the adequate resistance genes. The majority of the QTL identified usually show a minor effect on FHB resistance and have no diagnostic markers. Therefore, sustained and continuous efforts are still needed in cloning and validating resistance genes from hundreds of QTL associated with FHB resistance in wheat. The integration of various forward- and reverse-genetic approaches will be an important means to explore the genes of FHB resistance in wheat, with the development of plant–pathogen interaction mechanism in model plants ( Figure 1 ). In contrast to FHB resistance genes, modification of the susceptibility (S) genes will be an alternative option for controlling FHB. S factors or resistance suppressors have already been located on different chromosomes in wheat, yet, to date, they have not received much attention (Ma et al., 2006). In plants, some host factors encoded by S genes are always hijacked by pathogens through the secreted effectors to promote disease development. Mutations of these S genes evade the manipulation by pathogens and have been successfully utilized in crop disease control including wheat resistance to fungal pathogen (Garcia-Ruiz et al., 2021; Koseoglou et al., 2022). As secreted proteome of F. graminearum has been obtained, in depth analysis of the interaction between host and F. graminearum would facilitate the understanding of the function of S genes in wheat. Genes encoding resistance suppressors that always inhibit plant immune responses fall into another category of S genes (Garcia-Ruiz et al., 2021; Koseoglou et al., 2022). Mutations of these genes with abolished or reduced gene expression generally result in enhanced and durable disease resistance. These genes are significant components of crop disease resistance gene pool, while in FHB resistance, no such genes have been isolated so far. With the development of wheat mutant libraries, identification of such S genes will be facilitated. Though some genes with FHB resistance have been identified, the signaling pathway that wheat perceives and responds to F. graminearum attack remains obscure. Plant immunity involves large-scale changes in gene expression. Intensive investigation of the role of those genes responsive to F. graminearum in wheat FHB resistance will contribute to the cloning of practical resistance genes. With the completion of wheat genome sequencing, the initiation of pan-genomic research for tribe Triticeae, and the rapid development in biotechniques, breakthrough will be made in the field in the future.
HGM and HXM wrote the manuscript. YL, XZ and SZ collected some data. All authors contributed to the article and approved the submitted version.
This research was funded by the Jiangsu Key Project for the Research and Development (BE2022337) and the Seed Industry Revitalization Project of Jiangsu Province (JBGS2021047).
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647132 | Junpeng Wu,Yaxian Zong,Zhonghua Tu,Lichun Yang,Wei li,Zhengkun Cui,Ziyuan Hao,Huogen Li | Genome-wide identification of XTH genes in Liriodendron chinense and functional characterization of LcXTH21 | 27-10-2022 | Liriodendron chinense,XTH family,drought stress response,genome identification,root development | Liriodendron chinense is a relic tree species of the family Magnoliaceae with multiple uses in timber production, landscape decoration, and afforestation. L. chinense often experiences drought stress in arid areas. However, the molecular basis underlying the drought response of L. chinense remains unclear. Many studies have reported that the xyloglucan endotransglucosylase/hydrolase (XTH) family plays an important role in drought stress resistance. Hereby, to explore the drought resistance mechanism of L. chinense, we identify XTH genes on a genome-wide scale in L. chinense. A total of 27 XTH genes were identified in L. chinense, and these genes were classified into three subfamilies. Drought treatment and RT-qPCR analysis revealed that six LcXTH genes significantly responded to drought stress, especially LcXTH21. Hence, we cloned the LcXTH21 gene and overexpressed it in tobacco via gene transfer to analyze its function. The roots of transgenic plants were more developed than those of wild-type plants under different polyethylene glycol (PEG) concentration, and further RT-qPCR analysis showed that LcXTH21 highly expressed in root compared to aboveground organs, indicating that LcXTH21 may play a role in drought resistance through promoting root development. The results of this study provide new insights into the roles of LcXTH genes in the drought stress response. Our findings will also aid future studies of the molecular mechanisms by which LcXTH genes contribute to the drought response. | Genome-wide identification of XTH genes in Liriodendron chinense and functional characterization of LcXTH21
Liriodendron chinense is a relic tree species of the family Magnoliaceae with multiple uses in timber production, landscape decoration, and afforestation. L. chinense often experiences drought stress in arid areas. However, the molecular basis underlying the drought response of L. chinense remains unclear. Many studies have reported that the xyloglucan endotransglucosylase/hydrolase (XTH) family plays an important role in drought stress resistance. Hereby, to explore the drought resistance mechanism of L. chinense, we identify XTH genes on a genome-wide scale in L. chinense. A total of 27 XTH genes were identified in L. chinense, and these genes were classified into three subfamilies. Drought treatment and RT-qPCR analysis revealed that six LcXTH genes significantly responded to drought stress, especially LcXTH21. Hence, we cloned the LcXTH21 gene and overexpressed it in tobacco via gene transfer to analyze its function. The roots of transgenic plants were more developed than those of wild-type plants under different polyethylene glycol (PEG) concentration, and further RT-qPCR analysis showed that LcXTH21 highly expressed in root compared to aboveground organs, indicating that LcXTH21 may play a role in drought resistance through promoting root development. The results of this study provide new insights into the roles of LcXTH genes in the drought stress response. Our findings will also aid future studies of the molecular mechanisms by which LcXTH genes contribute to the drought response.
Plants are continuously exposed to various types of abiotic stress because they are sessile, and drought stress has a negative effect on the growth, yield, and cultivation of plants. Given that water scarcity reduces plant performance, improving the drought resistance of plants is a major goal of breeding efforts (Boyer, 1982; Bray et al., 2000; Cushman and Bohnert, 2000; Mittler, 2006). Plant improvement of drought resistance via molecular breeding approaches is a clear trend that further promotes the development of modern agriculture. With the various genome resources available, mining and utilizing genes that provide high resistance to drought stress will promote the development of plant molecular breeding (Parmar et al., 2017; Wai et al., 2020). In recent years, identification and functional analysis of various abiotic stress-responsive genes and transcription factors and their applications in breeding stress-tolerant plants were favored (Roy et al., 2011; Sharma et al., 2017). For example, overexpression of HhGRAS14 in Arabidopsis thaliana significantly improved the drought tolerance of transgenic plants (Ni et al., 2022). And orphan gene PpARDT was found to be involved in drought tolerance potentially by enhancing ABA response in Physcomitrium patens (Dong et al., 2022). MdFLP enhanced drought tolerance by regulating the expression of MdNAC019 in self-rooted apple stocks (Wang et al., 2022). The bZIP transcription factor ABP9 in maize is involved in the regulation of drought resistance, and overexpression of OsNAC10 in rice increased grain yield under drought stress (Jeong et al., 2010; Wang et al., 2017). The overexpression of a rice OsSalT in tobacco showed increased root growth and resulted in improved drought tolerance (Kaur et al., 2022). In Manihot esculenta, MeRSZ21b was found to be involved in drought tolerance, and plants overexpressed MeRSZ21b gene had longer roots than WT (Chen et al., 2022). Generally, there are gene families related to abiotic stress in plant genomes. The xyloglucan endotransglucosylase/hydrolase (XTH) family is a typical example (Eklöf and Brumer, 2010). XTH family belongs to the glycoside hydrolase 16 family (GH16), and XTH genes can be divided into four groups: I/II III-A, III-B, and the early diverging group (Campbell and Braam, 1999b). Increasing evidence has revealed that genes in the XTH family play an important role in drought stress. For example, plants overexpressing GmXTH23 had stronger drought tolerance and greater root lengths than wild-type (WT) plants (Long, 2020). Following overexpression of the hot pepper gene CaXTH3 in tomato, half of the stomata of transgenic plants were open under drought stress, and most of the stomata of WT plants were closed, a condition that was suggestive of transgenic plants having a higher drought tolerance than WT plants (Choi et al., 2011). XTH genes have also been shown to be involved in plant growth. For example, GUS staining of A. thaliana has shown that AtXTH genes might be expressed throughout all growth stages (Becnel et al., 2006). The expression levels of DcXTH2 and DcXTH3 increased dramatically during flowering, confirming that XTH genes play a role in petal growth (Harada et al., 2011). In addition, some plants use xyloglucan as a storage polysaccharide for embryonic development (Reid, 1985; Buckeridge et al., 2000). Due to their important roles in drought stress and growth, XTH family members have been identified in various plants, including A. thaliana (33 genes), Oryza sativa (29 genes), Hordeum vulgare (24 genes), Populus spp. (41 genes), Ananas comosus (24 genes), and Schima superba (34 genes) (Yokoyama and Nishitani, 2001; Yokoyama et al., 2004; Geisler-Lee et al., 2006; Fu et al., 2019; Li et al., 2019; Yang et al., 2022). L. chinense is a relict species in the Magnoliaceae family that has been widely used in timber production, landscape decoration, and afforestation. Drought stress has become a major barrier restricting the cultivation of L. chinense (He and Hao, 1999). As mentioned above, XTH family has been widely reported to be involved in drought response. However, the precise role and molecular mechanisms of the XTH family under drought stress remains unclear in L. chinense. Hereby, in order to have a better insight into the roles of XTH genes in L. chinense, we identified LcXTH genes on a genome-wide scale, uncovered LcXTH genes in relation to drought resistance and characterized their function. Overall, our study provides new insights into the possible roles of XTH genes in L. chinense and will aid future studies of drought resistance mechanisms in woody plants.
L. chinense material used in this study was obtained from Xiashu Forest Station at Nanjing Forestry University, Jurong, Jiangsu, China. In April 2021, leaves, roots and leaf buds were collected from a 30-year-old L. chinense tree originating form Songyang, Zhejiang Province, and immediately frozen in liquid nitrogen, and then stored at –80°C until further use. L. chinense seeds were soaked in water for 2 days, transplanted to soil, and cultivated in 40 cm × 30 cm × 4 cm trays for 2 months with a 16-h/8-h light/dark photoperiod. All L. chinense seedlings were watered twice a week. To investigate the response of L. chinense to drought stress, seedlings were watered with 10% PEG (100 g/L) solution. Samples of L. chinense were taken at 0, 3, 6, 12, 24, and 48 h and immediately frozen in liquid nitrogen for RT-qPCR analysis. All seedlings were grown under the same conditions, with the exception of plants in the drought stress treatment. Tobacco (Nicotiana benthamiana) was used for transgenic assays. Before sowing, seeds were sterilized with 75% (v/v) alcohol for 30 s, NaClO (v/v) for 15 min, and double-distilled water four times. Seeds were then sown on 1/2 MS medium (Murashige and Skoog, 2006) and vernalized at 4°C in the dark for 2 days. These seeds were placed in an incubator (SANYO, Japan) under a photoperiod 16-h/8-h light/dark photoperiod at 23°C for 10 days; they were then transplanted into MS medium and cultivated for 20 days. WT tobacco was used as a control, and all seedlings were grown in the same environment. The medium of the 30-day-old transgenic and WT tobacco seedlings was removed, and seedlings were planted in trays. Transgenic and WT tobacco plants were treated with 10% PEG (100 g/L) solution for 5 days, and root length was measured.
The two XTH domains (PF00722 and PF06955) were used as queries to search the L. chinense genome with HMMER (v. 3.0). The default settings and cutoff values were set to 0.001 (Potter et al., 2018; El-Gebali et al., 2019). Potential sequences were filtered using the Conserved Domain Search Service website (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi) (Marchler-Bauer and Bryant, 2004), and the candidate genes were identified using the SMART database (https://smart.embl.de/) (Letunic et al., 2021). Redundant genes were removed manually. The molecular weight (MW), isoelectric point (pI), and protein length were analyzed using the ExPASy website (https://web.expasy.org/protparam/) (Wilkins et al., 1999). Single peptides and the subcellular localization of LcXTH genes were predicted by SignalP (https://dtu.biolib.com/SignalP-6) and Plant-mPLoc (v. 2.0) (http://www.csbio.sjtu.edu.cn/bioinf/plant-multi/), respectively (Chou and Shen, 2010; Teufel et al., 2022). The Blast program was used to identify homologous genes (Chen et al., 2020).
Information on the chromosomal location of LcXTH genes was obtained from the genome GFF file. Ka and Ks values, protein similarity matrices, and collinear gene pairs were analyzed using TBtools (Chen et al., 2020).
A. thaliana proteins were downloaded from the TAIR website (https://www.A.thaliana.org/). H. vulgare proteins, O. sativa proteins were download from NCBI website (https://www.ncbi.nlm.nih.gov/). The sequences were aligned with ClustalW software (v. 2.1). A phylogenetic tree was constructed by MEGA 7.0 software using the neighbor-joining method with the following parameters: Poisson model, pairwise deletion, and 1000 bootstrap replicates (Kumar et al., 2016). The evolview website (https://www.evolgenius.info/evolview-v2/) was used to modify the phylogenetic tree (Subramanian et al., 2019).
The expression levels of LcXTH genes were evaluated using fragments per kilobase of transcript per million mapped reads (FPKM) values based on transcriptome data (https://www.ncbi.nlm.nih.gov/, PRJNA559687) from different tissues of L. chinense, and heat maps were constructed using TBtools (Chen et al., 2020). Gene Ontology (GO) analysis was performed using the clusterProfiler 4.0 (Wu et al., 2021).
The online software MEME (https://meme-suite.org/meme/doc/meme.html) was used to analyze the conserved motifs of XTH proteins with a maximum of 10 motifs (Bailey et al., 2015). The Gene Structure Display Server (http://gsds.gao-lab.org/) was used to analyze gene structure (Hu et al., 2014). The promoter sequence (2000 bp upstream of the start codon) of each LcXTH gene was extracted and then analyzed using PlantCARE online software (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/); the results were visualized using TBtools. The protein sequences were submitted to ESPript Web server (http://espript.ibcp.fr/ESPript/ESPript/) for secondary structure prediction.
A SteadyPure Plant RNA Extraction Kit (AG21019, Accurate Biotechnology, Hunan, Co., Ltd.) was used for RNA extraction following the user manual. The quality of RNA was assessed using a NanoDrop 2000 spectrophotometer. A260/A280 values ranged from 1.8 to 2.0, and values of A260/A230 ranged from 1.9 to 2.1; a total of 500 ng of RNA was used to synthesize complementary DNA (cDNA). RT-qPCR was used to analyze the expression profiles of LcXTH genes, and Actin97 was used as the reference housekeeping gene (Tu et al., 2019). The thermal cycling conditions for RT-qPCR were based on instructions provided in the SYBR Green Premix Pro Taq HS qPCR Kit (AG11701, Accurate Biotechnology, Hunan, Co., Ltd.). We used the 2−ΔΔCT method to calculate relative levels of expression (Livak and Schmittgen, 2001). Three biological replicates and technical replicates were conducted to ensure the accuracy of the results. And primers used in this study are listed in Table S1 .
The coding sequence of LcXTH21 was obtained from the genome of L. chinense, and primers were designed using Oligo software (v. 7). The full-length cDNA of LcXTH21 was amplified, and an 876-bp open reading frame sequence was obtained ( File S1 ). The cDNA of LcXTH21 was then cloned into the modified pBI-121 vector, which was digested with XbaI and BamHI QuickCut enzymes (Takara Biomedical Technology, Dalian, China). The transgene construct was introduced into Agrobacterium tumefaciens strain GV105, which was then transformed into tobacco using a leaf-disc infection method. We cut off the edge of wild-type tobacco leaves that had been cultured for about 30 days, put them in solid MS medium at 25°C for 2 days in the dark. Then we activated the transformed A. tumefaciens with liquid MS medium for 30 min to prepare an infection solution, then immersed the leaves in the infection solution for 10 min. After soaking, the leaves were cultivated continuously in the dark at 25°C for two days, then placed them at 25°C for 16 h in the light and 8 h in the dark. When callus grown on the edge of the leaves, we isolated callus and cultured on MS medium containing kanamycin. After 20 days of culture, PCR was used to determine whether the plants were positive, and positive plants were cultured until they reached maturity. The details can be seen in Figure S1 .
HMM searches were used to identify XTH genes. We originally obtained 29 putative XTH genes. These 29 candidate genes were then submitted to the SMART database, and incomplete sequences were removed manually. Finally, 27 XTH genes were obtained, which were named LcXTH1–LcXTH27. The characteristics and subcellular localization of LcXTH proteins were also predicted. The length of LcXTH proteins ranged from 243 to 337 amino acids, LcXTH26 and LcXTH27 were the largest proteins with 337 amino acids, and LcXTH17 was the smallest (243 aa). The theoretical pI values for LcXTH proteins ranged from 4.85 to 9.65, and the MW of these proteins ranged from 27.61 to 38.47 kDa. All LcXTH members were predicted to be localized to the cell wall, and 15 LcXTH members were predicted to be localized to the cytoplasm. Details are provided in Table S2 .
The chromosomal location of genes is determined by prior evolutionary events. We thus investigated the chromosomal locations of LcXTH genes. LcXTH genes were randomly distributed on nine chromosomes ( Figure S2 ). However, because the genome assembly was incomplete, the specific chromosomal locations could not be determined for two genes: LcXTH26 and LcXTH27. Most LcXTH genes were clustered on chromosomes 13, 14, and 17; chromosome 17 had 13 genes; and chromosomes 1, 3, 5, 9, and 16 had only one gene. Synteny within the LcXTH family was analyzed to clarify the evolutionary relationships among LcXTH genes, and three homologous pairs (LcXTH08-LcXTH12, LcXTH09-LcXTH14, and LcXTH02-LcXTH05) were identified ( Figure S2 ). The identity of LcXTH08-LcXTH12, LcXTH09-LcXTH14, and LcXTH02-LcXTH05 was 81.30%, 67.25%, and 82.8%, respectively. The substitution rates (Ka/Ks) of these three gene pairs were calculated to assess whether LcXTH genes have been subjected to selection ( Table S3 ). The substitution rates ranged from 0.099 to 0.114, which indicated that they have experienced purifying selection. These homologous pairs (LcXTH08-LcXTH12, LcXTH09-LcXTH14, and LcXTH02-LcXTH05) diverged approximately 66.42, 188.19, and 67.38 million years ago, respectively. We also evaluated the density of these genes across the entire genome. The density of these genes was high in regions with related homologous genes. Tandem duplication is an important mechanism underlying the expansion of gene families, and tandemly duplicated genes often occur in clusters (Kozak et al., 2009). Hence, we calculated the protein similarity matrix to investigate the identity of LcXTH genes in three gene clusters ( Figure S3 and Table S4 ). The identity of the LcXTH genes on chromosome 13 was 60.54%, and the identity of the three LcXTH genes on chromosome 14 ranged between 58.46% and 73.29%. Chromosome 17 contained 13 LcXTH genes, and the identity ranged from 47.16% to 96.59%. We speculate that the high identity of these closely arranged genes indicates that they are products of tandem duplication; generally, tandem duplication might be the major force driving the expansion of LcXTH genes.
To further investigate the evolutionary relationships among LcXTH family members, we constructed a phylogenetic tree using 113 XTH proteins. ( Figure 1 ). Phylogenetic analysis revealed that all XTH proteins were classified into four groups (group I/II, group III-A, group III-B, and the early diverging group). In L. chinense, most LcXTH genes (24) were categorized into group I/II. Only one gene (LcXTH03) was classified in group III-A. The remaining genes (LcXTH26 and LcXTH27) were grouped into III-B. No genes were classified into the early diverging group in L. chinense, which might stem from the incomplete genome annotation or gene loss.
To investigate the structure of LcXTH proteins, we submitted the protein sequences to the SMART database to identify conserved domains. We analyis the pylogenetic analysis of LcXTH proteins and two domains (Glyco_hydro_16 and XET_C) were identified in all LcXTH proteins ( Figure 2A and Figure 2B ). XET_C is a unique domain among GH16 family members, and the proteins in this family had a common structure known as the β-jellyroll fold (Atkinson et al., 2009; Eklöf and Brumer, 2010; Behar et al., 2018). We also performed conserved motif analysis on LcXTH proteins ( Figure 2C ) and found that the motif composition of LcXTH family members was similar. As shown in the schematic, the Glyco_hydro_16 domain included motifs 10, 6, 8, 3, 4, 2, and 1, and the XET_C domain included motifs 9, 5, and 7. Motifs 1, 3, 4 (ExDxE), and 5 were conserved in all LcXTH proteins. Previous studies have shown that the exon distribution of AtXTH genes is conserved within each subfamily in A. thaliana (Yokoyama and Nishitani, 2001; Yokoyama and Nishitani, 2008). We used the online tool GSDS 2.0 to analyze the exon–intron organization of the 27 LcXTH genes ( Figure 2D ). There were three or four exons in LcXTH genes, and the ExDxE domain was randomly distributed in these genes, with exception of the fourth exon. The signal peptides of LcXTH proteins were predicted. A total of 23 LcXTH proteins had signal peptides, and they were all located on the first exon. These short amino acid sequences might be responsible for transmembrane transport and have secretory functions.
To further characterize LcXTH proteins, two fully resolved structures of PttXET16-34 (PDB ID: 1UN1) and TmNXG1 (PDB ID: 2UWA) were used to characterize the secondary structures of XTH proteins with ESPript software. The schematic of the secondary structures shows that all LcXTH proteins contained the conserved ExDxE domain ( Figure 3 ). The first glutamic acid residue (E) acts as a catalyzed nucleophile, which typically initiates enzymatic reactions, and the second E residue acts as a base that activates the entering substrate (Fu et al., 2019). The N-glycosylation domain (NXT/S/Y) is thought to be critical for protein stability and is indicated in the figure. This N-glycosylation site was conserved in all group I/II proteins, but it was missing in nearly all group III-A XTH proteins (Eklöf and Brumer, 2010). However, the N-glycosylation site was observed in all LcXTH members, including LcXTH03 (a member of group III-A). The N-glycosylation sites of LcXTH proteins in group III were shifted by approximately 20 amino acids from the ExDxE domain to the C-terminus. The architecture of proteins was conserved within specific groups; for example, other conserved domains adjacent to the ExDxE domain were identified in LcXTH proteins, which were referred to as loop 1, loop 2, and loop 3. Previous studies have demonstrated that the extension of loop 2 plays a key role in determining the activity of XTH proteins (Baumann et al., 2007). In this study, Loop 2 was significantly shorter in groups I/II and III-B than in group III-A, suggesting that the difference in the length of loop 2 among subfamilies of L. chinense might partly account for the differences in the classification of these proteins and their functions. The sequence DWATRGG of loop 3 was present in most group I/II proteins; however, this sequence was replaced by SWATEN in group III-A members.
We analyzed the expression patterns of LcXTH genes across several tissues (including bracts, sepals, petals, stamens, pistils, leaves, and shoots) ( Figure 4A ). LcXTH genes showed tissue-specific expression patterns. LcXTH04, LcXTH12, LcXTH08, LcXTH25, and LcXTH26 were highly expressed in bracts, and the expression levels of these genes in other tissues were low. LcXTH07, LcXTH16, and LcXTH27 were significantly expressed in leaves. The expression levels of LcXTH03, LcXTH18, and LcXTH10 were high in pistils, suggesting that they are involved in pistil development. The expression levels of almost all LcXTH genes were low in sepals, petals, and stamens, suggesting that LcXTH genes were not expressed during flowering. The expression patterns of the three tandem arrays on chromosomes 13, 14, and 17 were not consistent. LcXTH13, LcXTH19, LcXTH20, LcXTH21, LcXTH22, and LcXTH24 were significantly expressed in shoots and had similar expression patterns, indicating that their functions might be redundant. However, the expression patterns of the other genes on chromosome 17 differed, suggesting that they might have acquired new functions. In addition, LcXTH01, LcXTH02, LcXTH11, and LcXTH17 were not expressed in any of the tissues examined in this study; this indicates that they are not expressed or have specific expression patterns that could not be detected in this study. GO analysis was performed to clarify the functions of LcXTH genes. All LcXTH genes encoded proteins with xyloglucan xyloglucosyl transferase and hydrolase activity, and they were all localized to the cell wall and the apoplast, which was consistent with the subcellular localization prediction ( Figure 4B ). All LcXTH genes were predicted to be involved in cellular glucan metabolic process and carbohydrate metabolic process.
To determine the expression patterns and regulatory characteristics of LcXTH genes, the 2000-bp upstream sequence of the translation initiation site (ATG) was extracted, and cis-elements were predicted ( Figure 5A ). A large number of cis-elements were involved in plant growth and development, stress responses, and phytohormone responses ( Figures 5B , C ). In the first category, the main cis-elements were motif CAT-box (33.64%), motif CCAAT-box (27.1%), motif O2-site (23.36%), and GCN4-motif (11.21%). Cis-elements involved in stress responses mainly included the ARE motif (40.78%), MBS motif (30.1%), LTR motif (14.56%), TC-rich repeats (7.77%), and GC-motif (5.83%). In the third category, the main cis-elements were abscisic acid-responsive element (ABRE, 58.48%), methyl jasmonate-responsiveness element (CGTCA, 14.8%), and auxin-responsive element (TGA-element, 7.97%). The most abundant phytohormone response element is associated with abscisic acid (ABA)-responsiveness, indicating that LcXTH genes might be regulated by ABA. To clarify the roles of LcXTH family members in drought resistance, we used BLASTp to identify stress-related LcXTH genes based on previous studies, and genes containing plant defense and stress elements (TC-rich motifs) were screened out (Xu et al., 2020). A total of eight LcXTH genes (LcXTH07, 15, 18, 19, 20, 21, 25, and 27) were identified, and their expression patterns under drought stress were clarified. The RT-qPCR results revealed that there were significant differences in the expression of these genes at different times under drought stress ( Figure 6 ). LcXTH07 expression was up-regulated at 0 and 6 h, down-regulated at 6 and 24 h, and highest at 48 h. Some genes exhibited similar expression patterns. For example, the expression patterns of LcXTH18, 19, 20, and 25 did not change significantly during 0 and 6 h, but significantly increased at 12 h. LcXTH21 exhibited the most rapid and strongest response to drought stress, and it was the most highly expressed gene. Some genes exhibited opposite expression patterns; the expression levels of LcXTH15 and LcXTH27 changed slightly, indicating that they might not play important roles in the response to drought stress.
In light of the strong response of LcXTH21 under drought stress, we cloned LcXTH21 and overexpressed it in tobacco to analyze its function; the characteristics of transgenic and WT tobacco plants were noted. Then, we cultivated transgenic plants and WT on 1/2 MS medium containing different PEG concentration. With the increase in PEG concentration, the root length decreased to varying degrees. Ten-day-old transgenic plants on 1/2 MS medium had an average root length of 1.26 cm, which was 68% longer than that of WT plants ( Figures 7A, B ). Under 5% PEG, the average root length in transgenic plants decreased by 52%, however, the WT root length decreased by 79% ( Figures 7C, D ). When the PEG concentration was increased to 10%, the average root length of transgenic tobacco and WT decreased by 95% and 154%, respectively ( Figures 7E, F ). Next, we treated thirty-day-old seedlings with 10% PEG for 5 days, and the root length were compared. The average root length was 2.11-fold that of WT plants ( Figures 7G, H ). To investigate the expression level between root and aboveground, we took samples of root and aboveground for RT-qPCR analysis. The results shown that the expression level of root was 42.06-fold than that of aboveground tissues/organs ( Figure 7I ). The plant hormone ABA plays a key role in regulating the resistance of plants to drought stress, whereas the NCED enzyme is a key rate-limiting ABA biosynthetic enzyme (Leung and Giraudat, 1998; Zhang et al., 2009; Zhu et al., 2017). Hence, to further study the relationship of the LcXTH21 genes and ABA signaling, we performed RT-qPCR analysis to detect the expression of NCED gene in thirty-day-old WT and transgenic tobacco ( Figure 7J ). The results showed that the relative expression level of NCED in transgenic plants was about 45-fold higher than that in WT controls, thereby suggesting that LcXTH21 might contribute to drought resistance by promoting ABA biosynthesis.
The evolution of the XTH family has received wide research interest. XTH genes were first detected in Zygnematophyceae and non-charophycean taxa; bacterial licheninases have long been considered non-plant ancestors of the XTH family because of their sequence similarities (Barbeyron et al., 1998; Michel et al., 2001; Del Bem and Vincentz, 2010; Behar et al., 2018). The expansion of XTH genes has also received much research interest. The relaxation of substrate specificity, including the broader specificity of group I/IImembers, might have contributed to the expansion of group I/II members (Shinohara and Nishitani, 2021). Tandem duplication is one of the important mechanisms underlying the expansion of XTH family members, and tandem duplications comprise 4.56% of the genome of L. chinense (Chen et al., 2019). We detected a large number of tandemly duplicated LcXTH genes, accounting for 66.7% of all LcXTH genes, indicating that tandem duplication is the main mechanism underlying the expansion of LcXTH family genes. All the tandem repeats identified in this study were group I/II members, indicating that tandem duplication has been a particularly important mechanism underlying the expansion of group I/II members. Previous studies have shown that some gene family members have gained new functions or undergone defunctionalization (Lynch and Conery, 2000). In this study, tandemly duplicated genes on chromosome 17 were of particular interest. The expression patterns of LcXTH13, 19, 20, 21, 22, and 24 were similar, and their functions might have been retained after genome duplication. The expression patterns of LcXTH16, 18, and 25 were inconsistent. The differential expression of these genes indicates that they might have gained new functions. LcXTH01, 02, 11, and 17 were not expressed in any tissues in this study, suggesting that they might have undergone defunctionalization. Understanding the classification of subfamilies is important for studying the evolution of the XTH gene family. Previous studies of monocotyledonous rice and dicotyledonous A. thaliana have shown that group I and group II members are difficult to distinguish; they have thus been usually classified into one group (group I/II) (Yokoyama et al., 2004). However, both sequence analysis and catalytic measurements have confirmed the divergence of group III (Fanutti et al., 1993; Yokoyama et al., 2004). In this study, a total of 27 LcXTH genes were identified, and they were grouped into three subfamilies. None of these genes were classified in the early diverging group, which might stem from an incomplete genome assembly or gene loss. Domain loop 2 in XTH family members plays an important role in subfamily classification, and previous studies have shown that the length of loop 2 contributes to the difference in the activity between XET and XEH members (Baumann et al., 2007). For example, the extension of loop 2 is thought to be the major structural change responsible for its endohydrolase activity in Fragaria vesca (Opazo et al., 2017). In this study, the extension of loop 2 was only observed in group III-A, and this might be an important factor contributing to the divergence of group III members. Glycosylation as one of key post-translational modifications can affect stability and the molecular weight of the target proteins (Ahmadizadeh et al., 2020; Heidari et al., 2020). In addition, proteins associated with the secretory pathway are firstly glycosylated in the endoplasmic reticulum. The above suggests that XET proteins may be involved in the secretory pathway. In this study, the N-glycosylation domain was observed in all LcXTH members, and previous research has demonstrated that the removal of this region significantly reduces the stability of some XET proteins (Campbell and Braam, 1999a; Kallas et al., 2005).
Stomatal closure is one of the first events that takes place in plants in response to drought stress to prevent water evaporation, and XTH genes play a crucial role in modifying the cell wall and altering its elongation under drought stress, which improves drought resistance (Kim et al., 2010). For example, overexpression of HvXTH1 in barley resulted in the enlargement of the stomata of transgenic plants relative to those of WT plants under drought treatment, indicating that transgenic plants were more drought tolerant (Fu, 2019). Overexpression of XTH genes from rose (Rosa rugosa) increased the drought tolerance of transgenic China rose plants (Chen et al., 2016). In addition, the presence of xyloglucan in early land plants suggests that XTH gene family members have played a key role in the transition from wetter to drier habitats (Popper, 2008). In our research, the drought treatment and RT-qPCR analysis showed that six LcXTH genes significantly responded to drought stress. The expression patterns of LcXTH18, 19, 20, and 25 were similar, suggesting that they exhibit similar responses to drought stress. ABA is an important signal in plants that mediates the response to drought stress. The most abundant phytohormone response element of LcXTH genes was associated with the ABA response (ABRE motif, 58.48%). Moreover, the augmented NCED expression levels detected in transgenic tobacco indicated that the increased drought stress resistance provided by the LcXTH21 transgene probably involved the promotion of ABA biosynthesis. The “balanced growth” hypothesis proposes that some plants can stimulate or maintain root growth while reducing shoot growth in response to drought stress (Bloom and Mooney, 1985). In Eucalyptus globulus, a drought-tolerant clone was found to have higher root growth rate than a drought-sensitive one (Costa et al., 2004). Moreover, the development of the root system largely determines the performance of plants under drought conditions. Thus, increased root biomass is one of the primary mechanisms used by plants to avoid, or reduce, drought stress (Kashiwagi et al., 2005). For example, the overexpression of OsNAC5 was found to enhance drought tolerance by increasing root diameter (Jeong et al., 2013). XTH gene family members have been widely studied considering the different roles they are known to play in root development. For example, seven XTH genes from rice were specifically expressed in the roots of seedlings (Yokoyama et al., 2004). AtXTH19 and AtXTH23 were involved in lateral root development via the BES1-dependent pathway, indicating that XTH genes play a role in root development (Xu et al., 2020). Later, the expression levels of AtXTH11, AtXTH29, and AtXTH33 in the roots and aboveground organs were found to differ in A. thaliana plants subjected to high temperature and drought stress, thereby suggesting that these genes might mediate rapid responses to drought stress (De Caroli et al., 2021). In a study performed in grapevine, the transcription levels of VvXTH genes presented the largest changes in roots and leaves under drought and salt stress, indicating that VvXTH genes vigorously respond to abiotic stress in leaves and roots (Qiao et al., 2022). Molecular breeding is a promising approach and great progress has been made in its use for improving the efficiency of plant breeding programs (Varshney et al., 2009). Combined with molecular marker-assisted selection, greater and faster genetic progress can be achieved. For example, in a study of Triticum aestivum, Iquebal et al. (2019) found that the maximum number of drought responsive quantitative trait loci were detected at the seedling stage and further analyzed the regulatory networks of key candidate genes and their roles in responding to drought stress in order to identify putative markers for breeding applications. Besides, in a recent study, Chauhan et al. (2022) used morphological, biochemical, and molecular markers from Withania somnifera to assess the 25 accessions of Indian ginseng, and concluded that these markers could be used to select superior ginseng genotypes. In addition, XTH genes were also reported to be involved in other abiotic stress, such as salt, heat and cold stress (Han et al., 2017; Hidvégi et al., 2020; De Caroli et al., 2021). Hence, we suggest that LcXTH genes play roles in drought resistance and other abiotic stresses. In this study, we performed functional characterization of LcXTH21, and transgenic tobacco showed higher drought resistance and more developed roots during seedling stage. Therefore, LcXTH genes could be potential functional markers when conducting marker-assisted-selection (MAS) for breeding varieties with high resistance to abiotic stress. As an example, we could select genotypes with high LcXTH21expression levels and these MAS genotypes could be expected to have higher resistance to drought stress at the adult stage. This procedure could represent an alternative way to breed L. chinense varieties with increased stress resistance associated with a highly developed root system in the coming decade.
In this study, 27 LcXTH genes were identified, and they were divided into three subfamilies. Tandem duplication was probably the major contributor to the expansion of the LcXTH family, and six LcXTH genes significantly responded to drought stress. Overexpression of LcXTH21 in tobacco resulted in a more developed root system. In summary, these findings enhance our understanding of the LcXTH gene family and lay the foundation for further exploration on drought resistance mechanisms in L. chinense.
The original contributions presented in the study are publicly available. This data can be found here: NCBI, PRJNA559687.
JW performed the experiments, analyzed the data and wrote the manuscript. YZ contributed to the data analysis and preparation of the manuscript. ZT took a role in experimental design and data analysis. ZH analyzed the data. LY, WL and ZC participated in plant sample collection and experimental assay. HL conceived the project, designed the experiments and revised the manuscript. All authors contributed to the article and approved the submitted version.
This study was financially supported by the National Natural Science Foundation of China (31770718 and 31470660) and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). The funding bodies played no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647133 | Maria Raffaella Ercolano,Daniela D’Esposito,Giuseppe Andolfo,Luigi Frusciante | Multilevel evolution shapes the function of NB-LRR encoding genes in plant innate immunity | 27-10-2022 | NBS-LRR genes,functional domain,genome organization,regulatory elements,plant receptor genes network,innate immunity | A sophisticated innate immune system based on diverse pathogen receptor genes (PRGs) evolved in the history of plant life. To reconstruct the direction and magnitude of evolutionary trajectories of a given gene family, it is critical to detect the ancestral signatures. The rearrangement of functional domains made up the diversification found in PRG repertoires. Structural rearrangement of ancient domains mediated the NB-LRR evolutionary path from an initial set of modular proteins. Events such as domain acquisition, sequence modification and temporary or stable associations are prominent among rapidly evolving innate immune receptors. Over time PRGs are continuously shaped by different forces to find their optimal arrangement along the genome. The immune system is controlled by a robust regulatory system that works at different scales. It is important to understand how the PRG interaction network can be adjusted to meet specific needs. The high plasticity of the innate immune system is based on a sophisticated functional architecture and multi-level control. Due to the complexity of interacting with diverse pathogens, multiple defense lines have been organized into interconnected groups. Genomic architecture, gene expression regulation and functional arrangement of PRGs allow the deployment of an appropriate innate immunity response. | Multilevel evolution shapes the function of NB-LRR encoding genes in plant innate immunity
A sophisticated innate immune system based on diverse pathogen receptor genes (PRGs) evolved in the history of plant life. To reconstruct the direction and magnitude of evolutionary trajectories of a given gene family, it is critical to detect the ancestral signatures. The rearrangement of functional domains made up the diversification found in PRG repertoires. Structural rearrangement of ancient domains mediated the NB-LRR evolutionary path from an initial set of modular proteins. Events such as domain acquisition, sequence modification and temporary or stable associations are prominent among rapidly evolving innate immune receptors. Over time PRGs are continuously shaped by different forces to find their optimal arrangement along the genome. The immune system is controlled by a robust regulatory system that works at different scales. It is important to understand how the PRG interaction network can be adjusted to meet specific needs. The high plasticity of the innate immune system is based on a sophisticated functional architecture and multi-level control. Due to the complexity of interacting with diverse pathogens, multiple defense lines have been organized into interconnected groups. Genomic architecture, gene expression regulation and functional arrangement of PRGs allow the deployment of an appropriate innate immunity response.
A large assortment of innate immunity receptors, able to perceive pathogen invasion and to activate a defense response, have evolved in plants. Cell surface receptors, such as receptor-like proteins (RLPs) and receptor-like kinases (RLKs), are mainly involved in monitoring the extracellular space to detect exogenous (microbe-associated molecular patterns, MAMPs) or endogenous elicitors (damage-associated molecular patterns, DAMPs) generated by plant pathogens (Heil and Land, 2014; Tanaka and Heil, 2021). A typical RLP structure is composed of an extracellular domain, responsible in MAMP/DAMP perception, a single-pass transmembrane region and a short cytoplasmic tail (Thomas et al., 1997). RLK are structurally similar to RLPs except for the presence of an intracellular kinase domain instead of the short cytoplasmic tail (Fritz-Laylin et al., 2005). Thanks to the kinase domain, RLKs are able to trigger signaling on their own (Liebrand et al., 2013 and 2014; Gust and Felix, 2014), whilst RLPs need to interact with a protein containing such domain to activate the downstream signaling (Jamieson et al., 2018). Intracellularly, nucleotide-binding leucine-rich repeat proteins (NB-LRRs or NLRs) can directly or indirectly recognize “effectors,” molecules secreted or delivered by pathogens into the cytoplasm to promote virulence (Van der Hoorn and Kamoun, 2008). NLRs have a stereotypical domain structure that allows them to recognize effectors and activate immunity. The nucleotide-binding (NB) domain, containing an ADP–ATP switch system that regulates the protein ON/OFF state, is the central module of NLR proteins (Takken et al., 2006). In addition, several leucine-rich repeats, that promote pathogen recognition and interact with the NB domain to prevent autoactivation, are found at the C-terminus (Wang et al., 2019). The N-terminal domain, thanks to the Toll/interleukin-1 receptor (TIR), coiled-coil (CC), resistance to powdery mildew 8 (RPW8) or similar domains, is mainly involved in downstream signal transduction (Bentham et al., 2017). The domain composition and architecture of pathogen receptor genes (PRGs) are important for the protein function. A domain is an evolutionary conserved entity because it has a specific functionality due to its fold. Thus, each conserved segment has a key role in protein function and folding (Moore et al., 2008). The two groups of PRGs share important characteristics and their activation promotes several common signaling pathways (Pruitt et al., 2021; Ngou et al., 2022). Both can trigger an immune response, including the activation of mitogen-activated protein kinase cascades, the production of reactive oxygen species, the increase in cytosolic calcium concentration and the expression of defense genes (Asai et al., 2008). In addition, NLRs can prompt a response that often culminates in a hypersensitive response (HR) (Dangl and Jones, 2001). It is worth noting that the RLP/RLK gene families result involved in several cellular processes, including growth, development and plant innate immunity (Andolfo et al., 2013), whilst NLRs are predominantly devoted to the activation of defense responses. The latter class showed a remarkable diversification within species and across species to meet specific needs (Barragan and Weigel, 2021). In addition, NLR gene signaling can rapidly augment the transcript and/or protein levels of key components of downstream immune response increasing the plasticity of innate immunity (Maruta et al., 2022). Plant innate immunity originated for combating diverse and ever-evolving pathogens and the complex organization of its main players has an important role in its functioning. Here, we summarize the current view of the dynamics of NLR domain arrangement and the genomic architecture of plant defense arsenal on different scales, ranging from physical organization to transcriptional regulation. We describe links between genome organization and various genomic processes, such as the interplay between different PRGs. Finally, we provide an overview of the multilevel organization of innate immune response.
The typical domains of NLR were already present in proteins of bacteria, protists, glaucophytes and red algae. In such organisms the NB is preferentially associated with domains like WD or beta-transducin repeats (WD40) or Tetratricopeptide repeat (TPR) domains to perform recognition/transduction activities (Andolfo et al., 2019). Several NB proteins with innovative domain combinations evolved in early plants. Independent NLR associations may have originated in Chlorophyta and in Charophyta algae (Sarris et al., 2016; Gao et al., 2018) by convergent evolution. An intriguing cross-species domain assembly between the NB domain and the LRR domain was highlighted in Charophyta unicellular green algae by Andolfo et al. (2019). The LRR-region of such genes showed high homology to RLPs, underpinning a putative cell-surface localization and an interconnected evolution history. Novel domain combinations have appeared, and the recombination of existing units has provided new functionalities. Best suited proteins with different cell locations from the plasma membrane (RLPs/RLKs) to cytoplasm (NLRs) have been employed for assembling a plant immunity network with the emergence of multicellularity. A burst of NB-LRR genes was observed in nonvascular plants (mosses, liverworts, and hornworts) mediated by reshuffling at the N- and C-terminal regions (Bornberg-Bauer et al., 2010; Sarris et al., 2016; Ortiz and Dodds, 2018). In vascular plants a large number of LRR encoding proteins, able to detect a variety of pathogens, was widespread in different species (Baggs et al., 2017). The structure and composition of such complex receptors have changed over time and the domain reorganization had an important role in evolutionary innovation (Urbach and Ausubel, 2017; Gao et al., 2018). The ancient domain remodeling was further complicated by functional links connecting domains, supradomains and multidomains during the evolution of domain organization (Nepomnyachiy et al., 2017; Aziz and Caetano-Anollés, 2021). Basic principles of PRG domain composition emerged by comparing the distributions of the theoretical and observed domain association in 33 eudicots, highlighting that the 30% of possible combinations were missed, more than 60% of protein showed two or three domains but up to 20% were single domains (Sanseverino and Ercolano, 2012). Favorable protein conformations can be promoted by specific domain combinations. In addition, proteins including domains with a two-component response, such as DNA-binding activity linked to transcriptional regulation of responses to stressors and signal transduction systems, may have some benefits (Aziz and Caetano-Anollés, 2021). The complex long-term coevolutionary arms race between plant and pathogens promoted species specific NLR combinations. For instance, TIR-NB-LRR (TNL) proteins are predominant in basal lineages and represent an important portion in the eudicot genomes but are absent in the monocots (Shao et al., 2016; Andolfo et al., 2017). A large reservoir of single domains or truncated NLR proteins is also scattered within resistance loci (Nishimura et al., 2017; Zhang et al., 2017). The evolutionary trajectories of plant receptor genes have been extended through the addition of endogenous and exogenous functional domains, such as the C-terminal jelly- roll/Ig-like domains (C-JIDs), found in many TNLs, that directly interact with effectors (Ma et al., 2020; Martin et al., 2020), or integrated decoy (ID) domains that can bind pathogen effectors (Cesari et al., 2014; Kroj et al., 2016). Unraveling protein architecture and discovering local sequence conservation and diversification provides the key to understanding how proteins evolve (Konagurthu et al., 2021). For instance, evolution studies on plant NB domain showed that motif patterns are rearranged to acquire more tuned functions and to refine folding ability (Andolfo et al., 2020). Over an evolutionary timescale, the immune receptors are under a strong selection pressure for fixing functional advantages.
Genomic-centric processes shaped the PRGs organization. NLR number can vary by orders of magnitude across different species with most plant genomes containing several hundred family members (Shao et al., 2016). Even closely related species can show lineage-specific mechanisms driving NLR expansions and contractions that reflect the plant lifestyle and the selection pressures derived from the environment (Tamborski and Krasileva, 2020), indicating that NLR evolution exhibits dynamic patterns of birth and death (Michelmore and Meyers, 1998). The NBS-LRR genes are not randomly distributed within plant genomes but rather are mainly organized in multi-gene clusters in hot-spot regions of plant genomes (Meyers et al., 1999; Hulbert et al., 2001; Richly et al., 2002; Zhou et al., 2004; Ameline-Torregrosa et al., 2008). NLR clusters can be divided into: homogenous clusters, including the same type of NLR, and heterogenous clusters containing diverse NLR classes. In addition, clusters containing a mixture of NLR, RLP and RLK were also found (Andolfo et al., 2013). Evolutionary forces governing gene clustering are not completely understood. The occurrence of gene duplication had great impact on expansion of this gene family in plant genomes ( Baggs et al., 2017). Copy number variation likely maintains a diverse array of genes to retain advantageous resistance specificities (Jiang and Assis, 2017). Non-functional copies can evolve into functional alleles conferring disease resistance and changes in a pseudogene can lead to the gain of function. Individual NLR genes have also been associated with extreme allelic diversity as a consequence of point mutations enriched in surface-exposed regions of LRRs for acquiring new pathogen recognition capabilities by positive selection, inter-allelic and paralog recombination and domain fusions (Joshi and Nayak, 2013). On the other hand, large genomic deletion/insertion can provide the loss/gain of a specific gene family. Plant species arsenals are set up by the interplay of large-scale gene organization, that determines global conservation in the order of loci, and extensive local genome rearrangements mediated by recombination, tandem duplication, segmental and ectopic duplication, unequal crossovers, transposons, horizontal transfer and other reshuffling elements (Andolfo et al., 2021). Adaptive diversification is induced by species-specific pathogen pressure thanks to the genome plasticity of plants (Mace et al., 2014; Di Donato et al., 2017; Mizuno et al., 2020). Regardless of the type of molecular mechanism, variations impact functionality and gene expression (Halter and Navarro, 2015). It seems that there is higher degree of association between genes in a cluster than just preferential co-localization. Recent studies showed that chromosomal regions with a defined gene density and activity, and with corresponding chromatin accessibility, histone modifications, and replication timing, are essential to orchestrate complex regulatory networks (Fritz et al., 2019). Each level of gene-genome intrinsic architecture is governed by mechanisms that we are just beginning to investigate (Nieri et al., 2017; Choi et al., 2018). An even more overwhelming challenge will be deciphering how PRGs are arranged, expressed and regulated within the three-dimensional (3-D) cellular context.
The plant immune response must be highly plastic and strictly regulated, given the different types of pathogens to counteract and its fitness cost. In fact, plants not challenged by pathogen show a basal level of PRG expression that is able to monitor non-self-mediated changes in the plant cell while minimizes costs of expression (Richard et al., 2018; Borrelli et al., 2018). On the other hand, when attacked by pathogens a tight regulation of the immune transcriptome is essential for the activation of effective defense response (Karasov et al., 2017). The plant immune transcriptome is regulated by many different, interconnected mechanisms that can determine the rate at which genes are transcribed. Epigenetic modifications, such as DNA methylation (associated with actively transcribed genes), are required in the regulation of PRGs. Trimethylation of lysine 4 of histone H3 (H3K4me3), di- or trimethylation of H3K36 have been identified being epigenetic modifications essential for the defense response (Richard et al., 2018). In addition, ubiquitination of histones regulates the expression of NLR genes (Zou et al., 2014; Lai and Eulgem, 2018). Interestingly, areas of the genome featuring NLRs also frequently contain high densities of transposons (Miyao et al., 2003; Wei et al., 2016; Lai and Eulgem, 2018), which may attract epigenetic modifications to reduce transcription in the area. TFs, such as WRKY, are involved in PRGs regulation through the binding to W-boxes found generally enriched in promoter regions of NLR genes (Mohr et al., 2010; Richard et al., 2018). Small RNAs (sRNAs), including microRNAs (miRNAs) and small interference RNAs (siRNAs) (Waheed et al., 2021) function as negative regulators of NLR transcripts (Zhai et al., 2011; Li et al., 2012; Shivaprasad et al., 2012; Halter and Navarro, 2015). The RNA surveillance pathways also have a leading role in the control of NLR-mediated resistance signaling. For example, nonsense-mediated mRNA decay (NMD) of NLR transcripts appears to play a role in defense induction similar to the miRNA/phasiRNA cascades. The perception of MAMPs can trigger transient suppression of NMD of NLR transcripts and, consequently, a temporary increase in NMD-targeted NLR transcripts, associated with enhanced disease resistance (Gloggnitzer et al., 2014). In addition, the nuclear RNA exosome regulates innate immunity in plants. For instance, mutations in components of the RNA exosome, which degrades RNAs in a 3′ to 5′ direction, suppress RPS6-dependent autoimmune phenotypes (Takagi et al., 2020). Alternative splicing can destabilize NLR transcripts triggering their own degradation and preventing their over accumulation (van Wersch et al., 2020). However, at least in some cases, alternative splicing can secure the synthesis of diverse transcript isoforms for full immunity (Jung et al., 2020). Recently published studies indicated that alternative polyadenylation (APA) of pre-mRNA is also an important regulatory mechanism of plant immune responses (Jia et al., 2017). APA can produce distinct transcript forms that differ in their coding sequences and in their 3’-untranslated regions, which are important for their function, stability, localization and translation efficiency of target RNA.
A complex network of interactions, based on intra- and inter-gene relationships, multilevel genome organization and DNA transcription and translation processes, regulates pathogen recognition events and defense responses. The defense mechanisms can be modulated through mutual interaction of a core set of receptors capable to activate the innate immunity responses (Saloman et al., 2020). The first discovered NLR-NLR cooperation dates to about twenty years ago, when it was discovered that two TNL genes, RPP2A and RPP2B, were required for resistance to downy mildew (Sinapidou et al., 2004). There are now many examples of NLR pairs, such as the Arabidopsis RPS4/RRS1 and the rice RGA4/RGA5 pairs (Cesari et al., 2014). One member (sensor) mimics the target of a pathogen effector, while the other member of the pair functions as a signaling ‘executor’ module that transduces the effector recognition. Moreover, it is emerging that many NLR-mediated immune responses require the presence and activity of so-called ‘helper’ NLRs, downstream signaling centers for a diverse array of sensor NLRs (Jubic et al., 2019). In this coupled reaction, sensor NLRs perceive effectors, and helper NLRs are involved in converting effector perception into immune activation (Cesari, 2018). Helpers are the Activated Disease Resistance 1 (ADR1), N Requirement Gene (NRG1) and NLR-REQUIRED FOR CELL DEATH (NRC1) (Gabriëls et al., 2007; Wu et al., 2016; Dong et al., 2016). Intriguingly, NRCs were first reported as required for the full function of transmembrane and cytoplasmic resistance receptors (Collier et al., 2011; Leibman-Markus et al., 2018). Functionally redundant NRC paralogs can display distinct specificities toward different sensor NLRs that confer immunity to oomycetes, bacteria, viruses, nematodes, and insects (Wu et al., 2017). The biochemical determinants that trigger helper-activation and physical interactions between sensor and helper remain unknown. Helpers could therefore act as ‘hubs’ to control signaling, guarding the whole immune signaling pathway rather than a specific molecule affected by an effector (Zhang et al., 2017; Leibman-Markus et al., 2018). Most likely, NLR helpers represent signal transduction and/or amplification levels that empower the innate immunity network (Wu et al., 2016). In addition, the plant pathogen immune response is promoted by the cooperation between the intracellular and extracellular receptors, even beyond early perception events (Yuan et al., 2021). A critical signaling component linking cell surface receptors and NLR-mediated immunity pathways is provided by reactive oxygen species produced by NADPH oxidase RBOHD (Yuan et al., 2021). High-throughput gene expression data can provide reliable information for the inference of PRGs (Calle García et al., 2022). In two tomato-pathogen-specific interactions, different networks of PRGs acting in concert were found (Andolfo et al., 2014). Although, plant immunity shares the same signaling mechanisms, the rewiring of PRGs networks may promote connection changes among defense pathways in specific plant-pathogen interactions. Investigation of differentially regulated PRGs could lead to the identification of pathogen-specific response patterns. Multiple responses can be merged into a single network model for capturing all the possible dynamic trajectories.
Defense scenarios can be depicted taking into account: the layer of defense, direct and indirect interaction, the network of response, cell sensing of pathogen and fitness needs (Andolfo and Ercolano, 2015). NLRs are involved in both perception and activation of immune signaling. Recent breakthroughs are starting to disclose mechanisms by which NLRs initiate immune signaling after effector perception. Conformational changes lead to the exchange of ADP by ATP and the oligomerization induction with the establishment of a functional ‘resistosome’ (Burdett et al., 2019). Complex formation, self-association or heteroligomerization was shown to be important for the activity of many NLRs (Casey et al., 2016; Zhang et al., 2017; Li et al., 2020; Jacob et al., 2022). Understanding how molecular entities evolve, work and are interconnected in any biological process is crucial. The high plasticity governing the innate immune system is founded on a complex functional architecture and a multi-level control as proposed in the Figure 1 model. Multiple levels, including gene structure, genome-gene relationships, gene regulation, molecular interaction show highly dynamic connections ( Figure 1 upper middle panel) that are able to regulate the innate immunity receptors with a different output, surveillance or defense response ( Figure 1 left and right panels). In particular, within this model, PRGs are involved in a “multi-actors” system, including NLRs that may act as sensors or “helpers” ( Figure 1 left and right panels). Leading sensors are able to coordinate a response, which may include the activity of different PRG groups (Andolfo et al., 2014; Ngou et al., 2021). The understanding of network structure, considering the distribution of the interaction strength, the challenges for the establishment of these interactions and the corresponding effects could be highlighted by a decomposition approach. It would be interesting to dissect the whole set of molecular interactions across the different levels, to identify the role and the spatial distribution of each element. Current knowledge of the immune network system is still limited and can be improved by studying its structural properties. Pathogen receptor system is continuously shaped over time to find its optimal arrangement thanks to different biological dynamics.
MRE, conceived the study and was primarily involved in writing the manuscript and in producing the figure. DD, substantially contributed to the writing and revising of the manuscript and was centrally involved in figure design. GA, contributed to the writing of immune response networking paragraph and revised the manuscript. LF, revised the manuscript and coordinated the work. All authors contributed to the article and approved the submitted version.
This work was supported by the Ministry of University and Research and carried out within the Harnesstom Project (101000716) funded by the European Community.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647139 | Daniel M. Freed,Josh Sommer,Nindo Punturi | Emerging target discovery and drug repurposing opportunities in chordoma | 27-10-2022 | chordoma,rare cancer,drug repurposing,target discovery,multi-omics,functional genomics,synthetic lethality,precision oncology | The development of effective and personalized treatment options for patients with rare cancers like chordoma is hampered by numerous challenges. Biomarker-guided repurposing of therapies approved in other indications remains the fastest path to redefining the treatment paradigm, but chordoma’s low mutation burden limits the impact of genomics in target discovery and precision oncology efforts. As our knowledge of oncogenic mechanisms across various malignancies has matured, it’s become increasingly clear that numerous properties of tumors transcend their genomes – leading to new and uncharted frontiers of therapeutic opportunity. In this review, we discuss how the implementation of cutting-edge tools and approaches is opening new windows into chordoma’s vulnerabilities. We also note how a convergence of emerging observations in chordoma and other cancers is leading to the identification and evaluation of new therapeutic hypotheses for this rare cancer. | Emerging target discovery and drug repurposing opportunities in chordoma
The development of effective and personalized treatment options for patients with rare cancers like chordoma is hampered by numerous challenges. Biomarker-guided repurposing of therapies approved in other indications remains the fastest path to redefining the treatment paradigm, but chordoma’s low mutation burden limits the impact of genomics in target discovery and precision oncology efforts. As our knowledge of oncogenic mechanisms across various malignancies has matured, it’s become increasingly clear that numerous properties of tumors transcend their genomes – leading to new and uncharted frontiers of therapeutic opportunity. In this review, we discuss how the implementation of cutting-edge tools and approaches is opening new windows into chordoma’s vulnerabilities. We also note how a convergence of emerging observations in chordoma and other cancers is leading to the identification and evaluation of new therapeutic hypotheses for this rare cancer.
Chordoma is an ultra-rare bone cancer that arises in the skull base or spine of pediatrics and adults, and originates from vestigial remnants of the embryonic notochord. Normally a low-grade but locally invasive disease, current standard of care for chordoma involves maximum surgical resection and/or radiotherapy (1). Despite significant advances in surgical techniques and radiotherapy strategies, the majority of chordoma patients eventually develop recurrent and/or metastatic disease and ultimately require systemic therapy to control further progression (2). To date, no drugs are approved for the treatment of advanced chordoma and conventional chemotherapy is generally ineffective (1, 2), resulting in a poor prognosis in the advanced disease setting. Research efforts over the past two decades have focused intensively on evaluating drug repurposing opportunities, primarily guided by detection of activated signaling pathways (3–5), focused drug screens (6–8), or anecdotal clinical responses to therapies (9, 10). These investigations inspired several Phase II clinical trials primarily involving multi-kinase inhibition with agents such as imatinib (11), sorafenib (12), lapatinib (13), or everolimus plus imatinib (14), for example, although modest efficacy and low objective response rates were observed in each study. In parallel, efforts to discover novel drug targets indicate that chordoma relies on the lineage-specific transcription factor brachyury (15–17), positioning it as arguably the most attractive – though, as of yet undruggable – target in chordoma. Over the same time period, the continued growth of genome-guided precision oncology prompted an explosion of drug repurposing efforts for molecularly-defined tumor types – a trend that also extended into the realm of rare cancers. For example, following its approval in chronic myelogenous leukemia, imatinib was successfully repurposed for KIT-mutant gastrointestinal stromal tumors (18), and dabrafenib plus trametinib was repositioned for BRAF V600-mutated anaplastic thyroid cancer (19) after the approval of this combination in non-small cell lung cancer (NSCLC) and melanoma. These and other success stories motivated a series of genomic profiling efforts in chordoma (20–27), with the hope that lifting the veil on chordoma genomes might reveal actionable therapeutic opportunities. Instead, these studies revealed that, similar to most sarcomas and pediatric cancers (28–30), chordoma appears to be characterized by a low and infrequently-actionable mutation burden – with only ~14% of chordomas harboring genomic biomarkers predictive of response to FDA-approved or investigational therapies in other indications ( Table 1 ) (31). Although this observation limits the current impact of traditional genomic profiling on drug repurposing campaigns and precision oncology efforts in chordoma, it does not mean that chordoma is devoid of exploitable alterations per se (32). Indeed, genomic profiling studies have identified several potentially actionable alterations based on emerging science – many of which we discuss further below – and validating these therapeutic opportunities may increase the number of advanced-stage chordoma patients that can benefit from genomics-guided precision oncology. Moreover, systematic functional studies in other rare cancers argue that multiple therapeutically actionable vulnerabilities nonetheless exist in the context of a genomically “quiet” background (33–35). In this review, we provide a snapshot of the emerging drug repurposing landscape in chordoma, while highlighting state-of-the-art approaches that can open new windows into chordoma biology to extend our view beyond that provided by genomics. We also discuss opportunities to repurpose lessons learned in other cancers to catalyze the identification of novel therapeutic hypotheses in chordoma. The synthesis of this emerging knowledge may lead to the discovery of new targets and the development of personalized drug repurposing opportunities for chordoma.
Although ~95-97% of chordomas belong to a single histological subtype, multiple observations suggest that its biology and disease mechanisms are heterogeneous. For example, over half of patients experience disease recurrence following complete tumor resection (2), and exhibit vastly different responses to systemic therapies in the advanced disease setting (36). Additionally, many chordomas are defined by complex genomic rearrangements (20, 37) or recurrent copy number losses (38), whereas other tumors harbor no detectable alterations. This molecular heterogeneity is also reflected in recent chordoma tumor profiling studies, which have utilized next-generation sequencing to identify potentially actionable alterations in chordoma ( Table 1 ) (20–23). These studies indicate that only ~14% of chordomas have biomarkers predictive of response to FDA-approved or investigational therapies in other indications. However, several opportunities for molecularly-guided drug repurposing are emerging based on recent scientific advances in chordoma and other cancers, and validation of these therapeutic hypotheses may increase the number of chordoma patients that can benefit from precision oncology ( Figure 1A ).
In one large cohort (20), PI3K pathway alterations were observed in 16% of cases (n = 17/104), indicating an opportunity to explore repurposing of inhibitors targeting PI3K or its downstream effector mTOR. One of the most frequently altered genes in chordoma is PTEN ( Figure 1A ); the resulting potential dependence on PI3Kβ signaling (40) suggests an opportunity to evaluate PI3Kβ inhibitors in chordoma (41). Indeed, a recent preclinical study revealed significant tumor growth inhibition by the pan-PI3K inhibitor buparlisib (BKM120) in patient-derived xenograft models (42). Downstream of PI3K, clinical trials involving mTOR inhibitor combinations have demonstrated modest clinical benefit in chordoma patients, particularly in tumors with mTOR effector activation (14, 43). Intriguingly, chordoma sometimes occurs in patients with tuberous sclerosis complex (44–46), which is characterized by loss of the mTOR negative regulators TSC1/2, further hinting at a role for the PI3K/mTOR pathway in chordoma pathogenesis. Moreover, PI3K and mTOR are regulated by receptor tyrosine kinases (RTKs), of which several appear to be activated in most chordoma tumors (4). Several studies have analyzed the activation state or effects of targeting RTKs including MET (47, 48), IGF1R (49, 50), and the FGFR family (51), though PDGFRβ (5, 10) and EGFR (3, 52) have received the most attention, primarily owing to evidence of some clinical benefit from agents targeting these RTKs (9–11, 13). Since RTK mutations are not frequently seen in chordoma, these receptors are presumably activated through alternative mechanisms such as aberrant growth factor production, which may be directly regulated by brachyury (53). The frequent activation of RTKs observed in chordoma may be related to the role of the notochord in regulating embryonic tissue patterning; in this context RTKs are thought to dictate proliferation and differentiation through the interpretation of morphogen gradients (54, 55). Inhibitors of wild-type EGFR, such as afatinib and cetuximab, have reproducibly shown promising activity against chordoma cell lines (3, 6, 7) and xenograft models (39, 47), which has motivated two Phase II clinical trials (NCT03083678 and NCT05041127). Since these strategies rely on inhibition of wild-type EGFR, it remains to be seen whether skin and gastrointestinal toxicities will limit their efficacy in the clinic (56). Growth factor signaling drives cell proliferation by upregulation of cyclin D, CDK4/6 activation, and progression through the G1/S cell cycle checkpoint. RTK activation along with frequent loss of the cell cycle tumor suppressor CDKN2A in chordomas (24, 38, 57, 58) has motivated preclinical repurposing studies with CDK4/6 inhibitors (39, 59, 60) and a Phase II trial involving palbociclib in CDKN2A-null chordoma patients (NCT03110744). It remains unclear, however, whether CDKN2A loss is a faithful predictor of sensitivity to CDK4/6 inhibition (61, 62) – possibly because, in addition to p16INK4A, CDKN2A encodes p14ARF, whose loss results in CDK2 deregulation and compensatory G1/S cell cycle progression. Nevertheless, tumors with co-deletion of the CDKN2A-proximal MTAP gene may present an opportunity for combinations involving CDK4/6 inhibitors and antagonists of the PRMT5 axis (63–65). Notably, CDK4/6 inhibition has been reported to potentiate T cell immunity in several contexts (66, 67), and we discuss opportunities for evaluating CDK4/6 inhibitor combinations in this context further below.
Genomic profiling studies have also revealed potential synthetic lethality strategies in chordoma. Several deleterious alterations have been reported in genes involved in DNA damage repair and response, including BRCA2, CHEK2, PALB2 and ATM (20, 23, 25, 26). In one recent study, a novel defective homologous recombination signature was identified in advanced chordomas that appears to impart a “BRCAness” phenotype and sensitivity to PARP inhibition (22). This strategy is being explored further in a Phase 2 clinical trial combining olaparib plus trabectedin for solid tumors with this defective homologous recombination signature (NCT03127215). Future studies aimed at examining the potential link between DNA damage repair defects and complex genomic rearrangements in chordoma may provide further mechanistic insight into this therapeutic opportunity.
Other studies have identified alterations in chromatin remodeling genes such as SETD2 and SWI/SNF complex members SMARCB1, ARID1A, and PBRM1 (20, 21, 24). Notably, biallelic loss of SMARCB1 defines an aggressive, poorly differentiated histopathological subtype of chordoma (<5% of cases) that most commonly afflicts the pediatric patient population (68, 69). Based on the apparent EZH2 dependence bestowed by SWI/SNF alterations (70, 71), a Phase II study is underway to explore repurposing of tazemetostat for SMARCB1-null chordoma (NCT02601950). The presence of SWI/SNF alterations also suggests opportunities for therapeutically exploiting aberrant SWI/SNF function, for example through resulting synthetic lethality with BRD9 antagonists (72–74), inhibitors of DNA repair (75–77), or p53 activation (78). Implementation of functional genomics screens may lead to the discovery of additional chordoma-specific synthetic lethal strategies in this context, which we discuss in more detail below. An interesting connection appears to exist between SWI/SNF, the Hippo pathway, and brachyury, chordoma’s main Achilles’ heel (79). Hippo transcriptional effectors YAP and TEAD are critical for notochord differentiation during embryonic development (80). Indeed, a YAP/TEAD motif is one of the top brachyury binding sites in chordoma cells (81), and reports have linked brachyury-mediated YAP upregulation to stemness and growth (82) – suggesting convergence between the Hippo and brachyury signaling networks. Intriguingly, SWI/SNF appears to sequester YAP, preventing its association with TEAD and thus antagonizing oncogenic Hippo transcriptional outputs (83). A key role of loss-of-function SWI/SNF alterations in chordoma may therefore be de-sequestration of YAP, which, when augmented by brachyury-mediated upregulation of YAP synthesis and stability, drives Hippo pathway flux to an oncogenic level. These observations suggest opportunities for evaluating an emerging class of TEAD palmitoylation inhibitors (84–86) in chordoma.
Although genomic profiling studies have informed our understanding of chordoma biology and expanded the list of potentially actionable therapeutic targets, chordoma nevertheless remains largely devoid of the recurring, actionable genomic alterations that define other solid tumors. For example, therapeutic biomarkers guide care for over two-thirds of metastatic NSCLC patients, with response rates to targeted therapies often approaching 70-80%, while chordoma profiling studies indicate ~14% of cases have potentially actionable genomic alterations ( Figure 1 ). As highlighted in the previous section, our developing understanding of cancer biology suggests up to an additional ~30% of chordomas might have actionable genomic alterations; nevertheless, a majority of advanced-stage patients lack clear or effective treatment options. This creates a need to open new windows into chordoma biology that extend our view beyond the “single oncogenic driver” perspective of cancer’s dependencies. To this end, studies across several cancers have revealed new categories of therapeutic targets, called “non-oncogene dependencies”, that mediate epigenetic changes, dysregulated signal transduction, metabolic rewiring, immune evasion, and other hallmarks of cancer (32). Multiple efforts are underway to analyze and integrate data layers derived from different aspects of cell biology, with a view to providing a more detailed molecular-resolution view of chordoma pathogenesis. For example, a recent investigation of methylation signatures in circulating tumor DNA revealed the existence of two distinct epigenetic subtypes in chordoma with prognostic relevance (87). A gene-set enrichment analysis pointed to dysregulated signaling pathways operating within each subtype, uncovering potential therapeutic opportunities that prompt further evaluation in functional studies. The exploration of additional data layers may further elucidate chordoma’s molecular subtypes, including their association with specific therapeutic vulnerabilities, risk of recurrence, and other features of the disease. Such multi-omics studies may also lead to the identification of tumor-specific or lineage-restricted cell surface proteins that can serve as targets for antibody-drug conjugates, bispecific antibodies, chimeric antigen receptor T cells, or other surface antigen-targeted modalities.
In addition to tumor cell intrinsic targets, therapeutic opportunities may exist within the tumor microenvironment, where crosstalk with various immune and stromal cell subsets can profoundly influence chordoma progression and therapy response (88, 89). Studies of the chordoma immune microenvironment to date have focused on the PD-1 axis (90, 91), as well as other potentially important immune checkpoints such as B7-H3 and HHLA2 (92, 93). A recent single-cell transcriptomic analysis of six chordoma tumors identified putative immunosuppressive contributions from regulatory T cells, tumor-associated macrophages, and TGFβ signaling (94). Notably, TGFβ pathway genes are upregulated by brachyury (81). These results point to a repurposing opportunity for antagonists of TGFβ signaling in combination with immune checkpoint blockade (95, 96). Interestingly, a chordoma patient treated with a bifunctional fusion protein targeting TGFβ and PD-L1 experienced late-onset tumor shrinkage in a Phase 1 trial (97). The set of factors that govern antitumor immunity is complex, and more comprehensive phenotyping of the chordoma immune microenvironment – through single-cell sequencing, digital spatial profiling, multispectral immunofluorescence and other approaches – will be important for creating an atlas of the various lineage states in chordoma and revealing therapeutically-reversible defects in the cancer-immunity cycle (98).
Other important tumor cell extrinsic features extend beyond the microenvironment, highlighting the need to study chordoma biology at various resolutions – including contributions from host physiology. For example, germline genetics are now understood to play a role in cancer predisposition (99) and tumor immunity (100). Additionally, the gut microbiome impacts immunotherapy efficacy in several solid tumor types (101–103), and recent data indicate that certain dietary habits can modulate the composition of the gut microbiome and influence immunotherapy response (104). Though it remains unclear how these factors contribute to the biology or treatment response of chordoma tumors, some studies are beginning to explore these questions. For example, MD Anderson’s Patient Mosaic initiative aims to collect genetic, immune, and microbiome profiles from thousands of cancer patients to inform treatment strategies. Biospecimens collected from chordoma patients enrolled on the cetuximab Phase II study at MD Anderson will be included in the Patient Mosaic protocol, shedding light on how host (and other tumor extrinsic) factors shape chordoma tumor biology.
The functional validation of new therapeutic targets and strategies resulting from multi-omics studies requires appropriate patient-derived samples and preclinical models. To this end, a variety of chordoma models have been developed by several groups (105, 106). In addition, the Chordoma Foundation has built a tumor biobank of over 500 biospecimens and a model repository currently consisting of 26 cell lines, 12 patient-derived xenograft (PDX) models, and a PBMC-humanized mouse model (www.chordoma.org/research). The majority of these models have been characterized by whole-exome and whole-transcriptome sequencing and will undergo additional multi-omics characterization in the future, with a view to facilitating hypothesis testing through the establishment of models representing the full diversity of chordoma. Moreover, these models are available to the research community, as are in-kind drug testing services offered through the Chordoma Foundation’s Drug Screening Program. As emerging drug repurposing concepts are evaluated in the Drug Screening Program, resulting data are publicly shared, whenever possible (107), to provide justification for further evaluation of the most promising therapeutic opportunities.
In translational cancer research, PDX models have been the gold standard for preclinical drug testing because they accurately recapitulate features of the patient’s tumor (108, 109); this has motivated the development of over two dozen chordoma PDXs by the Chordoma Foundation and others (47, 105) that represent the anatomical, age, histopathological, and known molecular diversity of chordoma. More recently, patient-derived organoids (PDOs) have generated significant interest as functional models because they provide faithful representations of patient tumors, while improving on initiation time, cost, and efficiency scales compared to PDXs (110). This technology is now being actively explored in chordoma; one recent proof-of-concept study reportedly developed chordoma PDOs from five different patients and screened them against various drugs to nominate personalized repurposing opportunities (111). In other cancer types, PDOs accurately mimic patient drug response (112–114) and have been utilized for personalized therapy (115, 116). The slow growth rate of chordoma tumors provides a large window of opportunity to develop protocols for establishing, validating, screening chordoma PDOs from high-risk or relapsing patients to enable identification of effective drug repurposing opportunities within the timeframe required to make treatment decisions.
Patient avatars like PDXs, PDOs and cell lines also serve as key platforms for target discovery because they allow functional studies capable of revealing or validating non-oncogene dependencies in chordoma. Genome-scale loss-of-function screens in various cancer cell lines have enabled the creation of “dependency maps” (33, 117), and this cutting-edge approach has recently been applied to chordoma to identify selective genetic dependencies (79). Perhaps unsurprisingly, T (or TBXT), the gene encoding brachyury, appears to be the most selectively essential gene in chordoma. Since brachyury (like most transcription factors) is a challenging drug target, the authors performed a drug repurposing screen and found that inhibitors of CDK9 or CDK7/12/13 (118) downregulate TBXT transcription and suppress chordoma cell proliferation. These results have motivated further in vivo testing of transcriptional CDK inhibitors, including KB-0742 (119), in the Chordoma Foundation’s Drug Screening Program (120). Ongoing systematic screening of genetic and chemical vulnerabilities in chordoma is facilitating the development of new therapeutic hypotheses. For example, CDK6 – but not CDK4 – appears to be a genetic essentiality in some chordoma cell lines (79). Outside of their common cell-cycle target RB1, CDK6 possesses a much broader substrate repertoire than does CDK4 (121) – suggesting that one or more non-RB1 targets may be mechanistically linked to chordoma’s CDK6 dependence. One interesting possibility relates to the observation that chordoma cells are sensitive to the lipid hydroperoxidase inhibitor RSL3 (79), which is known to promote ferroptotic cell death via antagonism of GPX4. CDK6 can upregulate glutathione and NADPH via phosphorylation of two glycolytic enzymes (122); depletion of these antioxidants can prime cells for ferroptosis (123). CDK6 may therefore be crucial for maintaining redox homeostasis in chordoma to safeguard against ferroptosis, providing rationale for evaluation of CDK4/6 inhibitors in combination with ferroptosis inducers. Chordoma’s apparent CDK6 dependence and potential ferroptosis susceptibility raises intriguing and unexpected parallels with clear-cell carcinomas (124, 125). Histologically, clear-cell renal cell carcinoma (ccRCC) is almost indistinguishable from chordoma, owing to morphological similarities between ccRCC’s characteristic lipid droplets and the physaliferous cells that define conventional chordoma (126). Notably, ferroptosis susceptibility in clear-cell carcinomas has been linked to HIF-1/2α-dependent accumulation of polyunsaturated lipids within the intracellular droplets that give rise to the clear-cell morphology (124). Both brachyury and mTOR are known to upregulate HIF-1α (81, 127–129), suggesting a possible connection between dysregulated hypoxia signaling, physaliferous morphology, and establishment of a ferroptosis-susceptible state in chordoma ( Figure 2A ). Although the precise composition of physaliferous vacuoles remains unclear (130–132), chordoma and ccRCC share additional similarities, including resistance to chemotherapy and modest mutational burdens enriched in chromatin modifier and PI3K/mTOR pathway alterations (133). Collectively, these observations suggest these cancers of different tissue origins share a similar cellular context, and potentially associated therapeutic vulnerabilities – providing opportunities for repurposing lessons learned from a well-studied and common cancer.
Another key implementation of systematic functional screens involves the discovery of synthetic lethal and combination therapy strategies. Loss-of-function screens in large cell line panels have led to identification of new synthetic lethal interactions (117, 134, 135), including PRMT5 dependence in cells with MTAP loss (63, 64). As noted above, the CDKN2A/MTAP locus is frequently deleted in chordoma (20, 21), suggesting an opportunity for repurposing PRMT5 or MAT2A inhibitors (136, 137). Exploiting such synthetic lethalities not only provides an avenue for targeting tumor suppressor loss in cancer, but is a particularly important approach to explore in genomically quiet malignancies. In addition to potential vulnerabilities created by loss of MTAP, SWI/SNF, or homologous recombination repair (as noted above), an intriguing candidate for synthetic lethality screening is LYST – a lysosomal trafficking protein of unknown function that’s lost in 10% of chordomas (20). Functional genomics screens in chordoma cell lines with LYST loss may reveal targetable vulnerabilities created by this unique alteration.
Preclinical and clinical research has yet to identify a therapy capable of producing frequent responses in chordoma (36), motivating the development of combination strategies aimed at increasing the magnitude and duration of therapeutic benefit. One approach with this goal in mind involves performing unbiased anchor screens, in which genome-wide CRISPR screening is utilized to identify genes whose loss sensitizes cells to a given targeted therapy ‘anchor’ (136). Such genes – if druggable – may serve as attractive targets for combination therapy regimens. A similar approach can also be employed to identify candidate resistance mechanisms – that is, genes whose loss (or gain) reduce sensitivity to the anchor drug. As one of the most well-validated therapeutic targets in chordoma, inhibitors of wild-type EGFR are arguably the best ‘anchors’ to initially explore in unbiased screens or rational combination studies. Indeed, combination therapy investigations with afatinib (47) or cetuximab (138) have yielded encouraging results. One interesting hypothesis involves combining cetuximab with a CDK4/6 inhibitor ( Figure 2B ). Since CDK4/6-mediated G1/S cell cycle progression is highly dependent on RTK/MAPK signaling, concomitant antagonism of EGFR-mediated cyclin D upregulation and CDK4/6 kinase activity may cause a more complete cell cycle arrest. This effect has been observed in lung (139) and pancreatic cancers (140), where combined MEK and CDK4/6 inhibition induced a profound G1/S arrest, resulting in a senescence-associated secretory phenotype (SASP) that promoted increased NK cell-mediated cytotoxicity and infiltration of CD8+ T cells, respectively. Importantly, as an IgG1 antibody, cetuximab monotherapy appears to promote antibody-dependent NK cell-mediated cytotoxicity in several cancers including chordoma (141). A cetuximab/CDK4/6 inhibitor combination may therefore act synergistically to halt the growth of chordoma tumor cells and provoke a strong NK- and T-cell based antitumor response. As a result, further exploration of this concept may be warranted, particularly once the single-agent activity of cetuximab (NCT05041127) and palbociclib (NCT03110744) in chordoma is benchmarked in the clinic. Notably, similar combination immunotherapy approaches aiming to enhance NK cell-mediated killing have recently been described in chordoma (138), and these strategies were reported to selectively target the reservoir of cancer stem-like cells that promote recurrence and therapy resistance.
Achieving deep and durable responses in chordoma will likely require identification of therapeutic concepts capable of invigorating antitumor immunity. Despite a low tumor mutational burden, a significant proportion of chordomas appear to be characterized by complex genomic rearrangements (20, 37), which may lead to high neoantigen expression. In addition to the examples noted above, documented patient responses to vaccines (142) and PD-1 inhibitors (143–146) provide important proof-of-concept for the use of immunotherapies in chordoma, and prompt evaluation of different immune checkpoints and combinations thereof. For example, strong scientific rationale exists for co-blockade of the PD-1 and TIGIT checkpoints in cancer (147), and a new clinical study enrolling chordoma patients is testing this concept with atezolizumab plus tiragolumab (NCT05286801). Another promising approach involves the cell-surface protein CD24, which is frequently expressed in chordomas and – along with brachyury and low molecular weight cytokeratins – has been used as a diagnostic marker for chordoma in some cases (148). Intriguingly, tumor-derived CD24 was recently identified as a key anti-phagocytic “don’t eat me” signal in other solid tumors (149), making it a promising immunotherapeutic target and prompting evaluation of CD24 blockade in chordoma. The evaluation of immunotherapy combinations in chordoma, such as PD-1 antagonism plus inhibition of TIGIT or TGFβ signaling as noted above, may reinvigorate the tumor-immunity cycle at multiple points. Multi-omics studies of chordoma may be valuable in guiding these efforts and revealing key molecular details governing the chordoma immune microenvironment.
Tumorigenesis appears to require three main ingredients: an oncogenic signaling input, deregulation of the signal through tumor suppressor loss (150, 151), and a permissive transcriptional environment for interpretation of oncogenic signaling (152, 153). Lineage-specific transcription factors – such as brachyury in chordoma – are essential for creating a permissive environment (79), and thus represent attractive drug targets. While oncogenic signaling and tumor suppressor loss can be targeted by kinase inhibitors and synthetic lethal strategies, respectively, transcription factors like brachyury are inherently challenging drug targets ( Figure 3A ). However, advances in drug discovery and the development of new targeted protein degradation technologies, such as proteolysis targeting chimeras (PROTACs) and molecular glues, provide opportunities to redefine this paradigm (154). To this end, numerous projects have recently been launched to develop novel compounds that bind brachyury with high affinity, which can either serve as functional inhibitors, molecular glues, or warheads for PROTACS. Notably, an open-source project through the Structural Genomics Consortium is focusing on the development of high-quality probes that bind pockets identified in brachyury crystal structures ( Figure 3B ) to induce industry investment in further brachyury drug discovery. The development of functional inhibitors is a challenging endeavor, given that brachyury lacks the deep binding pockets commonly associated with enzymatic activity. Yet, transcription factors like brachyury are often involved in multiprotein complexes, pointing to the development of compounds that modulate protein-protein interactions as an attractive strategy. For example, brachyury associates with the histone acetyltransferase p300 using an interface involving amino acid residue Y88 ( Figure 3A ) (155). The proximity of Y88 to residue G177 ( Figure 3A ) is interesting, as a G177D germline variant is strongly associated with chordoma (15). Because residue G177 is on a flexible, solvent-exposed loop, the G177D mutation is unlikely to affect brachyury structure – however this substitution may stabilize intermolecular contacts with p300 or other binding partners, thus modulating brachyury function. Indeed, the interaction between brachyury and p300 appears to regulate histone 3 lysine 27 acetylation (155) – a modification associated with active enhancers. Since association of brachyury with super-enhancers appears to be crucial to its role in chordoma (79, 81), designing compounds that can block or allosterically modulate this protein-protein interaction – for example, by targeting pocket A’ or F ( Figure 3B ) – may represent an attractive therapeutic strategy. Another novel approach to functionally modulating brachyury involves the development of Transcription Factor Targeting Chimeras (TRAFTACs) (156). In contrast to PROTACs, TRAFTACs utilize a transcription factor-specific DNA sequence to achieve target specificity, which is linked to an E3 ligase-recruiting moiety that directs brachyury to the proteasome for degradation. Although new drug discovery in an ultra-rare indication presents numerous challenges, the concept of “reverse” drug repurposing – that is, repurposing drugs initially developed in a rare cancer to more common indications – represents a promising path. Further highlighting the intriguing parallels between chordoma and kidney cancer, brachyury expression is associated with poor survival in ccRCC and papillary RCC ( Figure 4 ) (133, 157). Interestingly, expression of CDK6 – an apparent dependency in both chordoma and ccRCC (125), as noted above – appears to be regulated by brachyury (79). Numerous additional studies indicate brachyury is associated with poor prognosis and implicated in driving recurrence, metastasis, and/or resistance to standard of care therapy in several more common cancers including breast (158–161), lung (162–166), and colon (167). Thus, chordoma represents a “pure” and target-rich setting for the initial development of brachyury antagonists, which can then be expanded into larger indications where brachyury plays a role in disease progression.
Even when macroscopic complete resection is achieved using cutting-edge surgical approaches, the majority of chordoma patients experience disease recurrence and are unlikely to be cured (1, 2). At some point, local therapies such as surgery and/or radiation are no longer safe or feasible, and treatment options become limited due to a lack of effective systemic therapies. This has motivated intensive research to identify effective therapeutic strategies in chordoma, but drug repurposing efforts have been hampered by chordoma’s resistance to conventional chemotherapy and a paucity of actionable genomic alterations. In this review, we highlight several therapeutic hypotheses inspired by developing knowledge of chordoma biology and its parallels with other cancer types. In particular, we focus on emerging therapeutic opportunities based on emerging knowledge linking drug sensitivity to specific biomarkers. Nevertheless, through the lens of genomic sequencing, most chordomas still lack actionable alterations – underscoring the need to implement more sophisticated multi-omics approaches. Indeed, genomics is only one piece of the puzzle; tumor growth is controlled by multiple integrated systems, with each contributing uniquely to chordoma’s biology. Therapeutic opportunities exist within each of these systems, and efforts focused on elucidating and integrating them will provide a fuller view of chordoma’s biology. A key goal of multi-omics profiling efforts will be the identification of molecular subtypes, stratified by risk and therapeutic vulnerabilities, as has been demonstrated in other cancers (168–170). In parallel, unbiased functional assays, utilizing genome-wide CRISPR or high-throughput drug screening, may reveal non-oncogene dependencies or combination therapy strategies that would otherwise be difficult to detect through multi-omics profiling approaches. The identification of additive or synergistic therapeutic combinations is of particular interest, given the low historical response rates in chordoma (36). In addition to guiding target discovery campaigns, functional assays can provide personalized medicine opportunities. For example, if multi-omics studies identify patients at high risk for recurrence, the ability to establish and profile drug sensitivity of PDXs or PDOs at time of initial surgery may allow nomination of potential therapeutic options upon disease recurrence, as successfully demonstrated recently in breast cancer (116). Due to the intrinsically slow growth of chordoma tumors, such an approach could also be considered at the time of recurrence. Finally, we highlight how technological advances are opening the door to targeting the transcription factor brachyury, the main Achilles heel of chordoma. The identification of binding pockets on brachyury can serve as target sites for PROTAC warheads or molecular glues, but they may also be functionally important. One such potential site is pocket A’ or F ( Figure 3B ), near the putative p300 interface and residue G177, which is the site of a germline variant strongly associated with chordoma development. If efforts to target brachyury are ultimately successful, these drugs can be repurposed for more common cancers in circumstances where brachyury drives resistance to standard of care therapy.
All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.
This work was funded by the generous donors to Chordoma Foundation.
We acknowledge and thank our patient community for giving us constant motivation to help identify and develop better treatments for chordoma. We are also grateful for our generous donors who provide the necessary funding – and to our colleagues at Chordoma Foundation and the research community for their efforts – to work towards this goal. Finally, we thank Champions Oncology for their assistance with figure generation.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647155 | Jason P Mooney,Sophia M DonVito,Rivka Lim,Marianne Keith,Lia Pickles,Eleanor A Maguire,Tara Wagner-Gamble,Thomas Oldfield,Ana Bermejo Pariente,Ajoke M Ehimiyein,Adrian A Philbey,Christian Bottomley,Eleanor M Riley,Joanne Thompson | Intestinal inflammation and increased intestinal permeability in Plasmodium chabaudi AS infected mice [version 2; peer review: 2 approved] | 05-10-2022 | malaria,plasmodium,intestine,permeability,enteritis | Background: Gastrointestinal symptoms are commonly associated with acute Plasmodium spp infection. Malaria-associated enteritis may provide an opportunity for enteric pathogens to breach the intestinal mucosa, resulting in life-threatening systemic infections. Methods: To investigate whether intestinal pathology also occurs during infection with a murine model of mild and resolving malaria, C57BL/6J mice were inoculated with recently mosquito-transmitted Plasmodium chabaudi AS. At schizogony, intestinal tissues were collected for quantification and localisation of immune mediators and malaria parasites, by PCR and immunohistochemistry. Inflammatory proteins were measured in plasma and faeces and intestinal permeability was assessed by FITC-dextran translocation after oral administration. Results: Parasitaemia peaked at approx. 1.5% at day 9 and resolved by day 14, with mice experiencing significant and transient anaemia but no weight loss. Plasma IFNγ, TNFα and IL10 were significantly elevated during peak infection and quantitative RT-PCR of the intestine revealed a significant increase in transcripts for ifng and cxcl10. Histological analysis revealed parasites within blood vessels of both the submucosa and intestinal villi and evidence of mild crypt hyperplasia. In faeces, concentrations of the inflammatory marker lactoferrin were significantly raised on days 9 and 11 and FITC-dextran was detected in plasma on days 7 to 14. At day 11, plasma FITC-dextran concentration was significantly positively correlated with peripheral parasitemia and faecal lactoferrin concentration. Conclusions: In summary, using a relevant, attenuated model of malaria, we have found that acute infection is associated with intestinal inflammation and increased intestinal permeability. This model can now be used to explore the mechanisms of parasite-induced intestinal inflammation and to assess the impact of increased intestinal permeability on translocation of enteropathogens. | Intestinal inflammation and increased intestinal permeability in Plasmodium chabaudi AS infected mice [version 2; peer review: 2 approved]
Background: Gastrointestinal symptoms are commonly associated with acute Plasmodium spp infection. Malaria-associated enteritis may provide an opportunity for enteric pathogens to breach the intestinal mucosa, resulting in life-threatening systemic infections. Methods: To investigate whether intestinal pathology also occurs during infection with a murine model of mild and resolving malaria, C57BL/6J mice were inoculated with recently mosquito-transmitted Plasmodium chabaudi AS. At schizogony, intestinal tissues were collected for quantification and localisation of immune mediators and malaria parasites, by PCR and immunohistochemistry. Inflammatory proteins were measured in plasma and faeces and intestinal permeability was assessed by FITC-dextran translocation after oral administration. Results: Parasitaemia peaked at approx. 1.5% at day 9 and resolved by day 14, with mice experiencing significant and transient anaemia but no weight loss. Plasma IFNγ, TNFα and IL10 were significantly elevated during peak infection and quantitative RT-PCR of the intestine revealed a significant increase in transcripts for ifng and cxcl10. Histological analysis revealed parasites within blood vessels of both the submucosa and intestinal villi and evidence of mild crypt hyperplasia. In faeces, concentrations of the inflammatory marker lactoferrin were significantly raised on days 9 and 11 and FITC-dextran was detected in plasma on days 7 to 14. At day 11, plasma FITC-dextran concentration was significantly positively correlated with peripheral parasitemia and faecal lactoferrin concentration. Conclusions: In summary, using a relevant, attenuated model of malaria, we have found that acute infection is associated with intestinal inflammation and increased intestinal permeability. This model can now be used to explore the mechanisms of parasite-induced intestinal inflammation and to assess the impact of increased intestinal permeability on translocation of enteropathogens.
Symptomatic malaria parasite infection is characterised by a cyclical fever, anaemia, and malaise with Plasmodium spp-infected red blood cells (RBCs) detectable in the peripheral circulation. Gastrointestinal symptoms are commonly noted in malaria patients, with diarrhoea reported in both travelers and those residing in malaria-endemic areas (systematically reviewed, ( Sey et al., 2020)). For example, diarrhoea was reported in 25% of 451 Ugandan children hospitalised with malaria, significantly more frequently than among malaria-uninfected hospitalised children (11%) ( Lo Vecchio et al., 2021) and treatment with antimalarial drugs can resolve diarrhoeal symptoms within 48 hours ( Lo Vecchio et al., 2021; Sowunmi et al., 2000), suggesting, but not proving, a causal association. The primary role of the intestine is digestive, absorbing both water and nutrients whilst creating a barrier to invasion by microorganisms including pathogens. Disturbance of normal intestinal function can result in diarrhoea; a diverse clinical presentation being either watery (e.g. osmotic or secretory) or exudative (i.e. with mucus, blood and cellular discharge). Gastrointestinal pathogens, (protozoal, bacterial and viral) are highly prevalent in malaria-endemic areas and episodes of diarrhoea are common, especially among children ( Troeger et al., 2018). It can, therefore, be challenging to determine whether an episode of diarrhoea is caused by a concurrent malaria infection or is simply coincidental. One common intestinal pathogen, non-Typhoidal Salmonella (NTS), a particularly frequent cause of invasive bacterial disease (invasive NTS, iNTS) in sub-Saharan Africa resulting in considerable morbidity and mortality (reviewed ( Takem et al., 2014) is, however, significantly more common in people with, or recently recovered from, a clinical episode of malaria than among those with no recent history of malaria infection ( Biggs et al., 2014; Park et al., 2016; Scott et al., 2011). Whilst the features of diarrhoea associated with Plasmodium infection remain ill-defined ( Sey et al., 2020), an association with increased intestinal permeability ( Pongponratn et al., 1991) and decreased absorption of vitamin B 12 and D-xylose ( Karney & Tong, 1972) has been observed. Furthermore, autopsies of individuals dying from malaria have revealed intestinal haemorrhages ( Dudgeon & Clarke, 1919) and some evidence of sequestered infected RBCs in villous capillaries ( Pongponratn et al., 1991; Seydel et al., 2006). Given the continuing world-wide burden of clinical malaria ( Organization, 2021), the additional burden of subclinical malaria infections ( Stresman et al., 2020), and the burden of enteropathogenic infections ( Geus et al., 2019), it is important to understand associations between these infections at the intestinal level. To date, our understanding of malaria-associated intestinal disturbance comes largely from virulent murine models of severe and non-resolving Plasmodium infection. For example, Plasmodium yoelii infection has been associated with mild caecal inflammation, dysbiosis of the intestinal flora, increased colonisation with E. coli and NTS, and increased intestinal permeability (as measured by lactose:mannitol absorption ratios) ( Chau et al., 2013; Mooney et al., 2015). Small intestinal dysbiosis and pathology has also been observed in Plasmodium berghei ANKA infected mice ( Shimada et al., 2019; Taniguchi et al., 2015), and traditional serial-blood passaged Plasmodium chabaudi infection has been associated with increased cellular influx in the jejunum and increased intestinal permeability ( Alamer et al., 2019). Taking these findings together, a picture is emerging in which severe, acute Plasmodium spp. infection in mice induces intestinal inflammation leading to dysbiosis, increased intestinal permeability and increased colonisation by intestinal pathogens. However, the molecular and cellular processes underlying these intestinal responses, particularly whether they are driven by systemic or local inflammation, are unknown. These murine models of malaria are characterized by high parasitaemia and moderate to severe symptoms, and are therefore less representative of human malaria parasite infections in endemic settings. Therefore, to evaluate the intestinal response in a more physiologically-relevant model of mild to moderate malaria during acute and resolving infection, we have used the recently-transmitted model, in which mosquito transmission attenuates parasite virulence and modifies the host immune response ( Spence et al., 2013; Spence et al., 2015). Moreover, we used a fluorescently-tagged P. chabaudi AS line to facilitate imaging of infected RBCs to resolve whether intestinal inflammation is directly associated with parasite sequestration.
This study was reviewed and approved by the Ethical Review Body of the University of Edinburgh. All procedures were carried out in accordance with the UK Home Office regulations (Animals Scientific Procedures Act, 1986) under Project Licence number P04ABDCAA. Throughout this study, all efforts were made to reduce animal usage and ameliorate harm to animals. Mice were housed in the University of Edinburgh Licenced Animal Facilities 60/2605), and all animal procedures were performed in laboratories within the animal facilities. Six to eight weeks old female C57BL/6J mice were purchased from Charles River (Tranent, UK). All animals were maintained with at least one companion, randomly housed in individually ventilated cages furnished with autoclaved woodchip, fun tunnel and tissue paper at 21 ± 2°C, under a reverse light-dark cycle (light, 19.00 – 07.00; dark, 07.00 – 19.00) at a relative humidity of 55 ± 10% in a specified pathogen free facility. Mice were housed under these light-dark cycle conditions to allow collection of P. chabaudi trophozoites prior to schizogony at 13.00–15.00 hrs, and were allowed to adapt to a reverse-light schedule for at least 7 days before P. chabaudi infection. They were fed ad libitum an autoclaved dry rodent diet (RM3, Special Diets Services, UK), along with autoclaved water. Animals were monitored according to institutional guidelines, with routine daily health checks and increased monitoring during P. chabaudi infection. Euthanasia was performed by cervical dislocation at the end of phenotypic experiments, or by exsanguination under anaesthesia (pentobarbital sodium; Euthatal). This specific method of anaesthesia reduces animal suffering whilst maximising blood volume obtained.
The C57Bl/6 mus musculus- Plasmodium chabaudi chabaudi AS animal model of malaria was chosen to minimize host genetic variability and to obtain robust infections with a very low incidence of severe disease. Animals were inoculated i.p. with 1×10 5 PcAS-GFP or PcAS-mCherry-infected RBCs (iRBC) that had been blood passaged 3–6 times since primary infection by mosquito; deemed ‘recently mosquito-transmitted PcAS infection’, as previously described ( Spence et al., 2013). GFP or mCherry are constitutively expressed in the cytoplasm of these parasites at all stages of development ( Marr et al., 2020). In total, 208 mice were used in this study, in 8 experiments with groups of 4–7 mice, to provide statistical significance. Mice were infected with GFP-expressing PcAS (4 experiments, n=79), mCherry-expressing PcAS (4 experiments, n=51), or were uninfected controls (n=58). For each experiment, 2 mice were used to expand frozen stocks of stabilate parasites, (16 mice total). Four mice were excluded; three inoculated mice which were uninfected and one with an unexpectedly high parasitemia. For each experimental readout per time-point, two independent experiments were performed. Mice were weighed and monitored for haemoglobin concentration and parasitaemia by tail snip blood sampling at 18–21hrs of the blood-stage life-cycle for optimal detection of circulating trophozoites, as described previously ( Marr et al., 2020). Parasitemia was determined by flow cytometric analysis; diluting 1µL of tail blood in 1mL of Dulbecco's phosphate-buffered saline (dPBS, Gibco, UK) containing 5 IU mL -1 heparin sodium (L6510, VWR), and then diluting a further 1:5 prior to acquisition on a BD Fortessa (Becton Dickinson, UK). At least 100,000 events were analysed per sample; gates were set using uninfected control blood using FlowJo V10 (Tree Star), as previously shown ( Marr et al., 2020). Processing of control blood was performed prior to that from infected mice to minimise potential cross-contamination upon data acquisition. Haemoglobin concentration (Hb, g/L) was measured using a Hemocue Hb201+ (Radiometer, Sweden). Weight change was calculated as a proportion of an individual’s pre-infection weight, with measurements taken prior to tail snips. At various days post infection, mice were euthanized (at the time of predicted schizogony) by exsanguination under anaesthesia (pentobarbital sodium; Euthatal) for tissue and/or blood collection following cardiac puncture. Data are shown as ‘day post infection’. As blood sampling was timed to coincide with the presence of circulating trophozoites (i.e. before schizogony, which marks the completion of a replicative cycle) the number of completed replicative cycles is one less than the number of days pi.
Cardiac blood was collected into 5µL of heparin sodium (5 IU ml -1), centrifuged at 10,000g for 5 min, and plasma stored at -70°C for subsequent analysis. For multiplex analysis, a magnetic Luminex assay (LXSAMSM-7, R&D systems, UK) was performed according to the manufacturer’s instructions for IFNγ (BR33), TNFα (BR14), and IL-10 (BR28), using undiluted samples and analysed on a Bio-Plex 200 (Bio-Rad, USA). For IFNγ analysis by enzyme-linked immunosorbent assay (ELISA), plasma was diluted 1:2 and assayed with the mouse IFNγ ELISA MAX deluxe (430804, BioLegend, UK) according to the manufacturer’s instructions. Samples which gave values below the detectable range were reported at the limit of detection for each analyte.
At necropsy, the intestines were divided into five equal lengths (three for the small intestine, two for the large), and cleaned of contents by flushing with dPBS. Tissue was then immersed in 1mL RNAlater (Sigma-Aldrich, UK) and stored at -70°C after chilling according to the manufacturer‘s instructions. For isolation of RNA, tissue was transferred to 2mL FastPrep Lysing Matrix D tubes (MP Biomedicals) containing 1mL of TRIzol (Invitrogen). Tissues were then homogenized using a Precellys 24 tissue homogenizer (Bertin instruments) at 30sec on high speed, followed by phenol/chloroform extraction with TRIzol according to the manufacturer’s instructions. Residual DNA was then removed (AM1906, Ambion/Thermo-fisher). Purified RNA was measured using a NanoDrop spectrophotometer and diluted to 100 ng/mL prior to cDNA synthesis using the AffinityScript Multiple Temperature cDNA synthesis kit (200436, Agilent), according to the ‘1 st strand cDNA synthesis’ manufacturer’s protocol using 1µg RNA in a 40μL volume. For each sample, 2µL of the cDNA was transferred to a 96-well plate with 18µL of mastermix (dispensed by robot, Corbett CAS-1200) containing 10µL Brilliant III Sybrgreen Ultrafast Mastermix (600882, Agilent), 6.4µL ultrapure water (Gibco), and 0.8µL of both forward and reverse primers (diluted to 10µM) for each gene target (sequences listed in Table 1). Samples were run on a CFX96 Real-Time PCR Detection System (Bio-Rad, USA) at 96°C for 3min, followed by 40 cycles of 96°C for 5 sec and 60°C for 10 sec, and data acquisition. Data were analyzed using the comparative threshold cycle (C T) method. Target gene transcription of each sample was normalized to the respective levels of β-actin mRNA and represented as fold change over gene expression in control animals, as described previously ( Lokken et al., 2014). To summarise, to calculate the relative fold gene expression, an individual reference gene Ct value (β-Actin) is subtracted from the target gene Ct value (ΔCt), with the mean of control samples then subtracted (ΔΔCt), and finally the value is taken to two to the negative power (2- ΔΔCt).
Rolls of cleaned intestinal tissue were fixed immediately in 10% PBS-buffered formalin, followed by embedding into paraffin prior to sectioning into 4–5µm slices onto charged slides. Slides were deparaffinised and rehydrated using an AutoStainer XL (Leica) prior to staining. For visualisation of Plasmodium parasites, immunohistochemistry was performed targeting GFP in MT-PcAS-GFP-infected mice. Antigen retrieval was achieved by autoclaving (121°C, 45min) in TRS (pH 6.1; S169984-2, Agilent Dako), followed by washing in PBS/0.1% Tween20. Slides were blocked with 3% hydrogen peroxide for 10min, followed by non-specific horse serum matched to the secondary antibody for 15min, followed by blocking with avidin and biotin for 15min (927301, Bio Legend). Slides were incubated with goat anti-GFP (AF4240, R&D Systems) diluted 1:500 in PBS/0.1%Tween+1%FBS) in a humified chamber at 4°C overnight. After 3 x 15min washes in PBS/0.1%Tween, slides were incubated with biotinylated horse anti-goat IgG H+L (BA-9500, Vector Laboratories) diluted 1:500 for 1h at room temperature. Normal goat IgG (AB-108-AC, R&D Systems) or no primary antibody were used as controls. Finally, slides were stained with DAB (SK-4100, Vector Laboratories) using an ABC reagent kit (32020, Thermo Fisher), according to the manufacturers’ instructions with substrate development for 10min, and counterstained with hematoxylin (3136, Sigma Aldrich) using an autostainer XL (Leica). Villous height and crypt depth were measured on haematoxylin and eosin-stained sections scanned at 40x with a NanoZoomer (Hamamatsu Photonics, Japan) and analysed with QuPath Software (v0.2.3) ( Bankhead et al., 2017), an open platform for bioimage analysis. Quality assessment scoring (i.e. focus, small artefacts, orientation of the villi) was performed on randomized and blinded scans, followed by collection of 30 measurements of pairs of neighboring crypts and villi using the line tool. Each intestinal roll was divided into three sections (proximal, medial, and distal), with 10 crypts and villi measured in each area. Villus height to crypt depth ratio (Vh:Cd) for each neighbouring pair was calculated, then averaged for either the entire small intestine or each section.
Intestinal permeability was assessed as described previously ( Alamer et al., 2019; Denny et al., 2019; Taniguchi et al., 2015), with modifications. Food was withdrawn from cages for 5 hours prior to oral gavage with 0.1mL of 4-kDA fluorescein isothiocyanate (FITC) dextran (FD4-1g, Sigma) diluted to 25mg/mL in water, with the time of gavage recorded for each animal. Food was returned after gavage and mice were culled exactly 1 hour post gavage ( Volynets et al., 2016; Woting & Blaut, 2018). 100µL of plasma (collected as described above) was placed in a black, flat-bottomed 96-well plate, and fluorescence intensity at 520nm measured after excitation at 485nm in a FLUOstar Omega microplate reader (BMG Labtech). FITC-dextran concentrations were calculated from a standard curve of 10-fold serial dilutions of FITC-dextran standard and analysed using Microsoft Excel.
Large intestines were excised from anus to caecum, split open with scissors and the contents collected with a blunt metal edge into a 2mL eppendorf tube. Samples were placed at -70°C until processing. Contents were weighed and 0.5mL of ‘faecal buffer’ (0.5% anti-protease cocktail (P8340, Sigma) in dPBS) added. Samples were allowed to rest for 30min at 4°C, then placed on a vortex adapter for 30min with continuous shaking, as described previously ( Fidler et al., 2020). Faecal homogenates were centrifuged at 8,000g for 5min and 250μL of supernatant was stored at -70°C. Mouse proteins in faecal supernatants were enumerated by ELISA for IgA (88-50450, ThermoFisher), calprotectin (E1484Mo, Bioassay Technology Laboratory) and lactoferrin (EM1196, FineTest), according to the manufacturer’s instructions. Faecal supernatants were diluted 1:2 for lactoferrin detection and 1:400 for IgA, and were undiluted for calprotectin measurements. To measure residual FITC-dextran fluorescence, faecal supernatants were centrifuged a second time at 2,000g for 5min and then diluted 1:4 in water prior to reading at 485/520nm, as outlined above.
Markers of inflammation were compared between mice culled at 4, 7, 11 and 14 days post infection and uninfected control mice. Each post-infection group was compared with the control group using Dunnett’s test to account for multiple testing. Where necessary the data were log-transformed to improve symmetry and when there was evidence of heterogeneity in variance between the groups, Dunn’s test with Bonferroni adjustment for multiple testing was used instead of Dunnett’s test. Correlations between membrane permeability (FITC-Dextran concentration) and parasite load or fecal lactoferrin were assessed using Pearson’s correlation coefficient. All statistical analyses were performed, and graphs made, using GraphPad Prism (v 8.2.1 or v 9.1.0). A p value of <0.05 was considered statistically significant.
In mice infected with blood stage recently mosquito-transmitted P. chabaudi AS parasites, expressing GFP ( rMT-PcAS-GFP), parasitaemia peaked 9 days post infection (p.i.) at a low to moderate density (median 1.32%, IQR 0.53-4.27, n=51) ( Figure 1), in line with expectations ( Spence et al., 2013). There were no significant changes in weight compared to uninfected mice, but hemoglobin concentrations declined on day 11 p.i., as observed previously ( Marr et al., 2020). There was a clear but very transient inflammatory response on day 7 p.i. (i.e. immediately prior to peak parasitaemia) characterized by raised plasma concentrations of IFNɣ and TNFα. Plasma IL-10 concentrations also peaked on day 7 p.i. and were significantly raised for several days. These kinetics are typical of acute malaria parasite infections, denoting a switch from a pro- to anti-inflammatory systemic response to protect the host from immunopathology, driven in part by the co-production of IFNɣ+ and IL-10+ in activated T cells ( Couper et al., 2008a; Couper et al., 2008b; do Rosário et al., 2012). To determine whether rMT-PcAS-GFP infection and the associated systemic inflammatory response has any intestinal consequences, inflammatory markers were analysed in samples of duodenum, jejunum, ileum, caecum, proximal colon and distal colon by qRT-PCR ( Figure 2). Intestinal transcript levels for ifng and cxcl10 were raised between 7 and 11 days p.i., and were significantly higher than controls in all sections of the intestine on day 7 p.i. Raised inflammatory markers were particularly evident in the proximal colon with significant elevations of il10 and tnfa on day 7 p.i.; with lcn2, cxcl1, and mip2 at 11 days p.i. Whilst plasma IFNɣ peaks at 7 days p.i., intestinal transcripts of ifng remain elevated in the colon through to day 11 p.i., and show a significant positive correlation to the circulating protein levels in the proximal colon, but not the distal colon. Thus, ifng transcripts can remain in the intestine whilst circulating protein levels have become undetectable. These data are indicative of a generalised, low-grade enteritis which coincides with the period of peak parasitaemia and systemic inflammatory response. To determine whether parasite localization in the intestine may be driving the enteritis, qPCR for PcAS ribosomal 18S (r18s) was conducted on the same tissue samples ( Figure 3). PcAS r18s was detected in all sections of the intestine with the highest transcript levels detected on days 7 and 11 p.i. Mean intestinal PcAS r18s concentrations were highly correlated with peripheral parasitaemia on both day 7 p.i (r=0.84, p=0.008, n=8 from 2 independent experiments) and day 11 p.i. (r=0.99, p<0.0001, n=9 from 2 independent experiments). As mice had not been perfused to remove intravascular blood prior to dissection, it was possible that parasites detected in the intestine were simply circulating in blood. However, we could not rule out the possibility that parasites might be sequestered within blood vessels or had entered the tissues ( Brugat et al., 2014); cytoadherence of P. chabaudi AS via binding to the cell-surface receptor ICAM-1 has been reported in the spleen and liver ( Cunningham et al., 2017). Therefore, to determine the tissue localization of intestinal rMT-PcAS-GFP, formalin-fixed tissue sections were analysed by immunohistochemistry ( Figure 3). Parasites in the spleen were a mixed population of rings, trophozoites and schizonts, as reported previously during schizogony of P. chabaudi AS ( Brugat et al., 2014). By contrast, in the intestine, only trophozoites and ring forms were seen. Moreover, parasitised red cells in the intestine were clearly confined to intravascular spaces of mucosal blood vessels and smaller villous capillaries. These data suggest that, whilst iRBC circulate freely within intestinal vessels, there is no obvious indication of parasite sequestration in the intestine. Despite the lack of parasite sequestration in the intestine ( Figure 3), evidence of intestinal inflammation ( Figure 2) raised the possibility of morphological damage. Although there was no gross intestinal damage (villous epithelium was intact with no pathological evidence of leucocyte infiltration, haemorrhage or necrosis), there was a significant reduction in villous height/crypt depth ratio in the distal small intestine at days 7 and 11 p.i. ( Figure 4), indicative of mild villous atrophy and/or crypt hyperplasia ( Mills, 2019). However, neither villous height nor crypt depth alone was significantly different in the small intestine. Given the evidence of a generalized enteritis in PcAS-infected mice, we considered whether there might also be changes in intestinal permeability. For these experiments, mice were infected with recently mosquito-transmitted P. chabaudi AS expressing mCherry (rMT-PcAS-mCh) (in order not to interfere with measurement of FITC-dextran). To determine whether any changes in permeability might be linked to intestinal inflammation, faecal inflammatory protein concentrations were also assessed. Parasitaemia of rMT-PcAS-mCh peaked at a similar density to rMT-PcAS-GFP and between days 8–11 p.i. (median density 1.74%, IQR 0.86-4.67, n=21) ( Figure 5). Mice infected with rMT-PcAS-mCh were culled 1 hour after oral gavage with FITC-dextran ( Woting & Blaut, 2018), rather than after 4 hours as previously described ( Alamer et al., 2019; Denny et al., 2019; Taniguchi et al., 2015). Faecal homogenates (colon contents) were analysed for FITC-dextran, secretory IgA and two biomarkers of intestinal inflammation, calprotectin and lactoferrin; ( Lamb & Mansfield, 2011). No significant change was observed for sIgA and calprotectin concentrations. However, faecal lactoferrin concentrations were significantly above baseline on days 7, 9 and 11 p.i., rising 13.7-fold by day 11 (median 79 vs. 1160 pg/gram of colon contents). Furthermore, FITC-dextran concentrations were markedly and significantly lower in the faeces and higher in plasma 1 hour after oral administration on days 7 to 14 p.i. Of note, plasma FITC-dextran concentrations were highly correlated with peripheral parasite densities at day 7 p.i. (r=0.61, p=0.02) and day 11 p.i. (r=0.78, p=0.02), with plasma IFNɣ at day 9 p.i. (r=0.80, p<0.001), and with faecal lactoferrin concentrations on days 9 (r=0.66, p=0.05) and 11 p.i. (r=0.77, p=0.02) ( Figure 6). Taken together, these data suggest that both intestinal inflammation and increased intestinal permeability are secondary to circulating parasitaemia and associated systemic inflammatory response.
Using a rodent model of attenuated, resolving malaria (intraperitoneal injection of recently mosquito transmitted P. chabaudi AS) that more closely reflects mild to moderate human malaria infections, with rapidly resolving parasitaemia peaking below 2% and mild to moderate anaemia that resolves upon parasite clearance, we have confirmed previous reports of malaria-associated intestinal inflammation ( Alamer et al., 2019; Mooney et al., 2015; Shimada et al., 2019; Taniguchi et al., 2015) and significantly extended those observations. We have shown that parasite-iRBCs circulate freely in the intestinal vasculature but do not appear to sequester in this site; that the enteritis is generalized throughout the small and large intestines and coincident with the development and resolution of parasitaemia; and that intestinal permeability is markedly increased at the peak of parasitaemia and intestinal inflammation. This study has thus established a relevant murine model of malaria-associated enteritis that can be used to further our understanding of malarial disease and enteric co-infections. Previous studies have looked for sequestration of Plasmodium spp-infected erythrocytes in the intestines. Using a luciferase tagged line of P. chabaudi AS, Brugat et al. (2014) observed parasites in the liver, spleen and lung but little if any luminescence from whole intestinal tissues ( Brugat et al., 2014). P. falciparum-infected parasites have been identified in the small intestine, including the intestinal villi, at autopsy ( Pongponratn et al., 1991; Seydel et al., 2006) but the resolution of the images was insufficient to determine their precise anatomical localization. In our rMT-PcAS model, immunohistochemistry revealed that ring and trophozoite stage parasites were abundant in the mucosal and villous blood vessels but schizonts were not seen and there was no evidence of cytoadherence of iRBCs to the vascular endothelium or of infiltration of iRBCs into the extravascular spaces or deeper tissues. Previous studies using murine malaria models that lead to high parasitaemia, severe anaemia and significant weight loss ( P. yoelii ssp and serially blood-passaged P. chabaudi) have reported moderate intestinal inflammation, with infiltration of the intestinal mucosa by monocytes, mast cells, and T cells ( Alamer et al., 2019; Chau et al., 2013; Mooney et al., 2015), epithelial damage ( Mooney et al., 2015) and villous atrophy and haemorrhages ( Taniguchi et al., 2015). By contrast, rMT-PcAS induced enteritis was much more subtle with modestly reduced villous/crypt ratios but no gross epithelial damage, inflammation or haemorrhage. Overall, however, it seems that the severity of malaria-associated enteritis reflects the severity of the infection per se. This, taken together with the lack of evidence for sequestration of P. chabaudi-parasitised erythrocytes in the intestine and the close temporal correlation between enteritis and circulating parasite density (parasitaemia), suggests that the enteritis may be driven by systemic inflammation rather than localization of parasitised RBCs in the intestine. Although the enteritis observed during rMT-PcAS infection was relatively mild, it was sufficient to cause a marked increase in intestinal permeability during the period of peak parasitaemia, as evidenced by very rapid translocation of FITC-dextran from the gut lumen (colon contents) into the plasma. Moreover, this increased permeability was highly correlated with both parasitaemia and intestinal inflammation (faecal calprotectin), suggesting a causal pathway in which parasitaemia drives systemic inflammation, systemic inflammation drives enteric inflammation and enteric inflammation drives increased intestinal permeability. Increased intestinal permeability has been demonstrated in humans infected with P. falciparum ( Pongponratn et al., 1991) and in P. yoelii nigeriensis-infected mice ( Chau et al., 2013)) (in both cases using the lactulose mannitol test ( Fleming et al., 1990)) and in more P. berghei ANKA, Plasmodium yoelii 17XNL, and PcAS models (using FITC-dextran) ( Alamer et al., 2019; Denny et al., 2019; Taniguchi et al., 2015), with varying degrees of increased permeability observed either early or at the peak of infection. The consequences of this change in intestinal permeability, especially in terms of maintenance of the barrier function of the intestinal epithelium and risk of translocation, invasion and systemic dissemination of enteric pathogens such as NTS ( Mooney et al., 2019), remain to be fully discerned. This study reveals novel observational changes in the intestine during a mild, avirulent murine malaria parasite infection. Future work is needed to unpick the mechanism by which intestinal inflammation is induced, and how this relates to intestinal permeability. This would include phenotypical analysis of the cellular source of the inflammatory mediators observed (e.g. IFNγ, TNFα, lactoferrin), as well as their location within the tissue and their contribution to intestinal permeability. In summary, we have established a clinically relevant murine model of malaria-associated enteritis characterized by systemic and local inflammation and increased intestinal permeability. This model can be exploited to better understand the pathophysiology of enteric disease during malaria infections and to understand the mechanisms by which current or recent malaria infections substantially increase the risk of invasive enteric bacterial infections. |
|
PMC9647159 | Haijie Ma,Xinyue Meng,Kai Xu,Min Li,Fred G. Gmitter,Ningge Liu,Yunpeng Gai,Suya Huang,Min Wang,Min Wang,Nian Wang,Hairen Xu,Jinhua Liu,Xuepeng Sun,Shuo Duan | Highly efficient hairy root genetic transformation and applications in citrus | 27-10-2022 | citrus,genetic transformation,genome editing,Agrobacterium rhizogenes,virus | Highly efficient genetic transformation technology is greatly beneficial for crop gene function analysis and precision breeding. However, the most commonly used genetic transformation technology for woody plants, mediated by Agrobacterium tumefaciens, is time-consuming and inefficient, which limits its utility for gene function analysis. In this study, a simple, universal, and highly efficient genetic transformation technology mediated by A. rhizogenes K599 is described. This technology can be applied to multiple citrus genotypes, and only 2–8 weeks were required for the entire workflow. Genome-editing experiments were simultaneously conducted using 11 plasmids targeting different genomic positions and all corresponding transformants with the target knocked out were obtained, indicating that A. rhizogenes-mediated genome editing was highly efficient. In addition, the technology is advantageous for investigation of specific genes (such as ACD2) for obtaining “hard-to-get” transgenic root tissue. Furthermore, A. rhizogenes can be used for direct viral vector inoculation on citrus bypassing the requirement for virion enrichment in tobacco, which facilitates virus-induced gene silencing and virus-mediated gene expression. In summary, we established a highly efficient genetic transformation technology bypassing tissue culture in citrus that can be used for genome editing, gene overexpression, and virus-mediated gene function analysis. We anticipate that by reducing the cost, required workload, experimental period, and other technical obstacles, this genetic transformation technology will be a valuable tool for routine investigation of endogenous and exogenous genes in citrus. | Highly efficient hairy root genetic transformation and applications in citrus
Highly efficient genetic transformation technology is greatly beneficial for crop gene function analysis and precision breeding. However, the most commonly used genetic transformation technology for woody plants, mediated by Agrobacterium tumefaciens, is time-consuming and inefficient, which limits its utility for gene function analysis. In this study, a simple, universal, and highly efficient genetic transformation technology mediated by A. rhizogenes K599 is described. This technology can be applied to multiple citrus genotypes, and only 2–8 weeks were required for the entire workflow. Genome-editing experiments were simultaneously conducted using 11 plasmids targeting different genomic positions and all corresponding transformants with the target knocked out were obtained, indicating that A. rhizogenes-mediated genome editing was highly efficient. In addition, the technology is advantageous for investigation of specific genes (such as ACD2) for obtaining “hard-to-get” transgenic root tissue. Furthermore, A. rhizogenes can be used for direct viral vector inoculation on citrus bypassing the requirement for virion enrichment in tobacco, which facilitates virus-induced gene silencing and virus-mediated gene expression. In summary, we established a highly efficient genetic transformation technology bypassing tissue culture in citrus that can be used for genome editing, gene overexpression, and virus-mediated gene function analysis. We anticipate that by reducing the cost, required workload, experimental period, and other technical obstacles, this genetic transformation technology will be a valuable tool for routine investigation of endogenous and exogenous genes in citrus.
Citrus is among the most important fruit crops worldwide and is grown in more than 114 countries (Talon and Gmitter, 2008). Global predicted citrus production exceeded 146 million tons (FAOSTAT; https://www.fao.org/faostat/en). The citrus industry currently requires new cultivars with desirable traits to improve yields, nutritional value, and adaptability to biotic and abiotic stresses. The application of genetic transformation to improve citrus has increased in recent years (Peña, 2000; Boscariol et al., 2006; Fagoaga et al., 2006; Febres et al., 2008; Gambino and Gribaudo, 2012). Agrobacterium-mediated transformation of epicotyl segments requires tissue culture, which is widely employed to produce disease-resistant materials in the laboratory, remains the quickest method for improvement of citrus cultivars. To date, many important traits have been successfully introduced into different citrus species and hybrids, such as lime, sweet orange, and grapefruit (Domínguez et al., 2002; Fu et al., 2011; Jia et al., 2017; Peng et al., 2017; Jia et al., 2021). However, transformation mediated by Agrobacterium tumefaciens has many disadvantages, including that the procedure is time consuming, laborious, expensive, and inefficient. The low frequency of rooting is an additional limitation of A. tumefaciens-mediated transformation of epicotyl segments of citrus (Gutiérrez-E. et al., 1997). As a result, micrografting is frequently used for maintenance of transgenic plants (Poles et al., 2020). Therefore, a rapid and highly efficient genetic transformation method bypassing the need for tissue culture is critical for gene function analysis and genetic improvement of citrus. Agrobacterium species are widely used to generate transgenic plants as the agrobacteria can integrate transfer DNA (T-DNA) into a host plant’s nuclear DNA genome. Agrobacterium tumefaciens transfers the tumor-inducing (Ti) plasmid into the host nucleus to incorporate exogenous DNA into a host chromosome and subsequently cause formation of a tumor at the plant wound site. This mechanism has been utilized for A. tumefaciens-mediated plant transformation of many plant species to improve crop traits and for research on gene function (Chetty et al., 2013; Kaur and Sah, 2014). In recent decades, A. tumefaciens has been widely applied in citrus breeding and gene functional research (Domínguez et al., 2002; Stover et al., 2013; Hongge et al., 2019). However, for most plants, especially woody species, when using A. tumefaciens, the generation of stable transformants requires plant regeneration from a few cells or even a single cell using exogenous phytohormones, and thus the process is time consuming and laborious. In addition, Agrobacterium rhizogenes has been successfully used in plant genetic transformation technologies (Estrada-Navarrete et al., 2007). Agrobacterium rhizogenes can infect plants to induce formation of hairy roots from wounded tissue owing to the expression of rol genes encoded in the Ri plasmid. The T-DNA cassette from the exogenous binary vector can be transferred and integrated into the host cell genome together with T-DNA from the Ri plasmid (White et al., 1985). Compared with A. tumefaciens, A. rhizogenes-mediated hairy root genetic transformation technology bypassing the requirement for tissue culture and antibiotic screening is highly efficient and has been widely used in many herbaceous plants for rhizosphere physiology research and recombinant protein production (Tsuro et al., 2005; Majumdar et al., 2011; Habibi et al., 2016; Beigmohamadi et al., 2019). However, A. rhizogenes-mediated genetic transformation in woody plants bypassing tissue culture remains at an early stage of application (Irigoyen et al., 2020). In this study, we describe a rapid and highly efficient root genetic transformation and genome-editing protocol for citrus using A. rhizogenes strain K599. This technique requires only 2–8 weeks for completion, bypasses tissue culture, and has applicability for diverse citrus accessions. To date, no report is available on a highly efficient endogenous gene-editing technology for citrus that bypasses tissue culture. In addition, the proposed protocol can be used for viral vector inoculation bypassing tobacco-mediated virion enrichment, which can improve the efficiency of virus-mediated analysis of citrus gene function. We anticipate that the protocol will be a valuable tool for routine investigation of endogenous and exogenous genes in citrus.
The Escherichia coli (DH5α) competent cells (CAT#: DL1002), A. rhizogenes (K599) competent cells (CAT#: AC1080), and A. tumefaciens (EHA105) competent cells (CAT#: AC1012) were obtained from Shanghai Weidi Biotechnology Co., Ltd. All transformed bacterial strains were stored in 15% glycerol and preserved in a freezer at −80°C. Escherichia coli cells were cultured in lysogeny broth medium at 37°C. The K599 and EHA105 strains were recovered and cultured at 28°C in tryptone yeast medium with corresponding antibiotics. Branches of Citrus medica, C. limon, C. sinensis, and citrange ’Carrizo’ were obtained from the National Citrus Engineering Research Center, Chongqing, China. Plants with transgenic hairy root were grown in a greenhouse at 26°C with a 16 h/8 h (light/dark) photoperiod. All citrus plants were cultured in a net greenhouse under natural conditions.
Recombinant A. rhizogenes strains were cultured in fresh yeast extract peptone medium with appropriate antibiotics at 28°C. The resuspended A. rhizogenes K599 cells at the final concentration (OD600 = 0.6) were diluted into the MES solution (10 mM MgCl2, 10 mM MES [pH 5.6], and 200 µM AS). Blade-removed citrus branches (approximately 2 months old) were collected from the greenhouse. We cut the stems into ~10 cm sections using sterilized shears by keeping the smooth surface of the cross-section. The base of the stems sections was soaked in the A. rhizogenes K599 suspension and vacuum infiltrated for approximately 25 min using a standard vacuum. The stem sections were cultured in a dome tray filled with vermiculite-mixed soil in the greenhouse at 26°C with 90% relative humidity and a 16 h/8 h (light/dark) photoperiod. Hairy root development began after approximately 2–4 weeks (C. medica) or 4–8 weeks (C. limon and citrange ‘Carrizo’) after agroinfiltrated transformation. Potential transgenic roots were detected by the fluorescence signal with a portable excitation lamp (Luyor-3415RG, Shanghai, China). The fluorescence-positive hairy roots were incubated in liquid modified Hoagland’s nutrient medium (Coolaber, Beijing, China) in sterile tubes for observation of symptoms. Transgenic roots were cultured at 26°C under a 16 h/8 h (light/dark) photoperiod. The symptoms were captured with a digital camera (Canon EOS 200D, Tokyo, Japan). Those shoots were further confirmed by PCR analysis. The hairy root transformation efficiency was calculated using the following formula: [(Number of GFP-containing roots)/(Total number of roots)] × 100. The GFP fluorescence in transgenic citrus roots was observed with a confocal microscope (LSM 780, Carl Zeiss, Jena, Germany) with 488 nm excitation and 505–530 nm emission wavelengths.
Genomic DNA from roots was extracted using the cetyltrimethylammonium bromide method (Richards et al., 1994). Total RNA from the root and callus was extracted using the RNA Isolater Total RNA Extraction Reagent (R401-01, Vazyme, Nanjing, China). Gel electrophoresis and a NanoDrop spectrophotometer (NanoDrop Technologies, Inc., Wilmington, DE, USA) were used to assess RNA quantity and quality. The cDNA was synthesized using the HiScript III 1st Strand cDNA Synthesis Kit (+gDNA wiper) (R312-01, Vazyme). RT-qPCR analysis was conducted using the AceQ Universal SYBR qPCR Master Mix (Q511-02, Vazyme) and a real-time PCR system (Q2000A, LongGene, Hangzhou, China). The internal control gene used was actin and the corresponding primers were listed in Table S1 .
The RNA-sequencing experiments were conducted using three biological replicates of each sample. Sequencing libraries were generated using the NEBNext Ultra II RNA Library Prep Kit (New England Biolabs, Ipswich, MA, USA) and were sequenced on an Illumina NovaSeq 6000 Sequencing System in paired-end mode. The raw reads were processed with Trimmomatic v. 0.36 (Bolger et al., 2014) to remove adaptor sequences and low-quality reads. The cleaned reads were aligned to the Citrus medica genome using HISAT2 v. 2.2.1 (Kim et al., 2015). The number of reads mapped to each gene was counted with htseq-count v. 1.99.2. Differential expression between the C. medica wild type and C. medica agroinfiltrated explants was analyzed using the ‘DESeq2’ R package (Love et al., 2014). Genes with an adjusted p-value (p adj ≤ 0.05) and at least two-fold change in expression were assigned as DEGs. To explore the functions and pathways of the DEGs, GO terms and KEGG pathway enrichment analyses were performed using the ‘ClusterProfiler’ R package (Wu et al., 2021) and were visualized using the ‘ggplot2’ R package. The selected KEGG pathways associated with “plant hormone signal transduction”, “amino acid biosynthesis”, “plant–pathogen interaction”, and “MAPK signaling pathway” were visualized using the ‘Pathview’ R package (Weijun and Cory).
All transgenic roots and the wild-type plants were subjected to PCR (P111-01, Vazyme) using gene-specific primers ( Table S1 ) to amplify DNA insertions or fragments including the target sites. The PCR amplicons were cloned into the pGEM®-T Easy vector (Promega, Madison, WI, USA) for Sanger sequencing. The sequence chromatograms were analyzed with SnapGene software.
Transgenic roots were first confirmed with a hand-held excitation lamp (Luyor-3415RG). For microscopic inspection, roots were rinsed with ddH2O and photographed under a Leica-M205FA stereomicroscope (Leica Microsystems, Wetzlar, Germany). Two different autofluorescence emission wavelength bands were used for detection: green (505–550 nm) and red (>560 nm), defined by optical filters. Transverse sections of the transgenic roots were cut with a razor blade as thinly as possible. The sections were mounted on a glass microscope slide in ddH2O. The GFP signal was observed under a Leica-SP8MP confocal fluorescence microscope (Leica Microsystems). The fluorescence was observed under excitation at 488 nm and emission at 505–550 nm. Simultaneously, images (1920 × 1024 pixels) were captured with a suitable scale bar.
The GUS (uidA) gene expression was detected using a GUS gene quantitative detection kit (SL7161, Coolaber) following the manufacturer’s instructions. Briefly, roots and hairy roots were incubated in 0.1 M sodium phosphate buffer with GUS substrate for 12 h at 37°C. The enzymatic reaction was stopped with 70% ethanol. Tissues were observed with a light stereomicroscope after GUS staining.
sgRNAs applied in this study were designed using the online tools CRISPRP v2.0 (http://crispr.hzau.edu.cn/cgi-bin/CRISPR2/CRISPR) based on their evaluation score (Rank from high to low), GC content (40%-60%) and putative off-target sites (Rank from low to high). target sequences with 20-bp were designed for each gene in the first exon. The purpose of this design will increase the possibility of affecting protein function and the likelihood that at least one site would be edited.
Prepare the A. rhizogene strain K599 harboring corresponding CLBV-based vector. Grow the corresponding A. rhizogene strain in YEP medium till OD600 = 0.8. Spin down the pellets and wash twice with infiltration medium (MES medium: 10 mM MES, 10 mM MgCl2, 200 µm acetosyringone). Re-suspend the pellets using infiltration medium till OD600 = 0.5. Place the infiltration medium at 25°C for 3 hours in dark condition. Conduct vacuum agroinfiltration using citrus explants and maintain explants in greenhouse at 25°C. Observe the phenotypes after several months and confirm the transcription of target genes by RT-qPCR. All the experiments were conducted using at least three replicates.
All datasets supporting the conclusions of this article are included in the article and supplementary files. The transcriptome project has been deposited at NCBI BioProject under the accession PRJNA800116 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA800116). All gene sequences, genome, gff3 file, and proteomes, gene ontology (GO), COG category, CAZy, KEGG, PFAMs, NR, eggNOG annotation of C. medica for bioinformatic analysis are available in the Zenodo repository at https://doi.org/10.5281/zenodo.5902607. The high-resolution figures are available on the figshare repository: https://doi.org/10.6084/m9.figshare.19060802.v1. The raw sequence data used for transcriptome analysis are available in NCBI under the Sequence Read Archive (SRA) with the SRA accession number for C. medica GFP agroinfiltrated explants: SRR17731820, SRR17731819, SRR17731818; C. medica wild-type: SRR17731817, SRR17731816, SRR17731815.
Agrobacterium rhizogenes integrates T-DNA from the Ri plasmid into the host plant genome when sensing signal substances, such as acetosyringone (AS), so as to induce formation of hairy roots and to synthesize substances needed for growth of the bacteria ( Figure S1 ). To achieve genetic transformation bypassing tissue culture, the percentage explant survival of 22 citrus genotypes in vermiculite was first analyzed, ranging from 0% to 95% at 45 days post-incubation ( Table S2 ). The four citrus genotypes with the highest percentage survival (>80%) were Citrus medica, C. limon, C. grandis ‘Shatianyou’, and C. hystrix. ( Figure 1A ). Most branches of these four genotypes eventually produced roots for more than 10 genotypes, the percentage survival was less than 40% within 45 days and most of the branches failed to produce roots ( Figure 1B ).
Based on the percentage survival, we selected C. medica, C. limon, and citrange ‘Carrizo’ for assessment of the efficiency of genetic transformation of citrus branches mediated by A. rhizogenes K599 harboring a binary plasmid (1380-Cas9-HA) overexpressing GFP ( Figure S2 ; Table S3 ). Briefly, the genetic transformation protocol comprised three steps: K599 and explant preparation, vacuum infiltration, and explant incubation in vermiculite ( Figure 2A ; Table S4 ). The fluorescent transgenic hairy roots began to develop 2 weeks (C. medica) or 4 weeks (C. limon and citrange ‘Carrizo’) post-incubation in vermiculite ( Figure S3 ). The length and number of fluorescent hairy roots increased significantly after 4 months ( Figure 2B ). Non-transgenic hairy roots lacking GFP signal also emerged, but these roots had no impact on subsequent research as they were easily distinguishable based on GFP fluorescence and were readily removed. Confocal microscopic examination confirmed that GFP fluorescence was universally distributed in transgenic hairy roots but was not observed in non-transgenic hairy roots ( Figure 2C ). Subsequently, genetic transformation in citrange ‘Carrizo’ was performed using a different binary plasmid carrying a gene encoding protein β-glucuronidase (GUS) ( Table S3 ). Staining of GUS revealed that dark blue-stained transgenic hairy roots were induced by K599 harboring the corresponding plasmid, whereas no GUS staining was detected in the control ( Figure 2D ). The transformation rate of corresponding plasmids or citrus species was listed in Table S5 . Compared with A. tumefaciens-mediated citrus genetic transformation, which usually takes 3–6 months with a low success rate, the K599-mediated genetic transformation method was time-saving and cost-effective.
To verify the efficiency of K599-mediated genetic transformation and genome editing, transformation was performed using 18 transformed K599 strains, each containing a different plasmid. All 18 plasmids carried the GFP and Neo encoding genes, and 11 contained genome-editing elements targeting different loci in the citrus genome ( Figure 3A ; Figure S4 ; Tables S3 and S6 ). Branches of C. medica were vacuum infiltrated using these transformed K599 strains. After one month, transgenic hairy roots corresponding to these 18 plasmids were obtained. Owing to the presence of the GFP selection marker, non-transgenic hairy roots were easily detected and removed, and thus C. medica with only transgenic hairy roots was obtained ( Figure 3B ). The GFP and Neo genes were both amplified from genomic DNA isolated from the transgenic hairy roots ( Figure S5 ). Sequencing revealed that the gRNA target in the corresponding transgenic hairy roots had been successfully edited ( Figure 3C ). The missing nucleotides in edited transformants were mainly located at the 5’ end of the PAM site, which is consistent with the properties of Cas9-mediated DNA cleavage. To detect possible chimeras in the transgenic hairy root tissue, primers ( Table S1 ) were designed to amplify the edited loci based on the sequencing results. No fragments were amplified by PCR from the edited hairy roots, and specific fragments of the expected length were amplified from wild-type explants ( Figure 3D ), confirming the absence of chimeras in the edited hairy roots. Furthermore, PCR amplification and sequencing were performed on one randomly selected edited hairy root (gRNA10) using primers that flanked the gRNA-targeted locus ( Table S1 ). A total of 20 random colonies were sequenced, which all contained changed sequences at the gRNA targeted locus, further confirming that the tested transformant was not a chimera ( Figure S6 ).
Compared with the wild-type explant lacking fluorescence, several GFP fluorescence spots were observed in callus formed at the wound surface of C. medica explants at 10 days post-agroinfiltration ( Figure 4A ), indicating that K599 successfully transferred the binary plasmid into numerous cells. Transcriptome sequencing (PRJNA800116) of callus from wild-type and agroinfiltrated explants was conducted. A total of 4,233 differentially expressed genes (DEGs), comprising 2,744 upregulated and 1,489 downregulated genes, were identified in agroinfiltrated tissues compared with the non-treated control ( Figure 4B ; Table S7 ). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis revealed that many DEGs in agroinfiltrated tissue were enriched in the “transcription factors” (87), “plant–pathogen interaction (55)”, “amino sugar and nucleotide sugar metabolism” (41), “plant hormone signal transduction” (56), “transporters” (136), “starch and sucrose metabolism” (37), “glycosyltransferases” (50), “cytochrome P450” (32), and “enzymes with EC numbers” (66) ( Figure 4C ; Table S8 ). Gene ontology (GO) enrichment analysis revealed that many DEGs in agroinfiltrated tissue were enriched in “DNA-binding transcription factor activity” (274), “transcription regulator activity” (290), “response to external stimulus” (326), “response to oxygen-containing compound” (347), “response to other organism” (243), “response to external biotic stimulus” (243), “response to biotic stimulus (243), and “defense response” (212) ( Figure 4D ; Table S9 ). Further analysis revealed that many DEGs enriched in the pathways “Plant hormone signal transduction” ( Figure S7 ), “Phenylalanine, tyrosine and tryptophan biosynthesis” ( Figure S8 ), “Plant–pathogen interaction” ( Figure S9 ), and “MAPK signaling pathway” ( Figure S10 ) were upregulated.
Huanglongbing is among the most destructive citrus diseases worldwide and is caused by the phloem-limited bacterium Candidatus Liberibacter asiaticus (CLas). Citrus roots harbor the pathogens during CLas infection and are difficult to treat with bactericides. Our previous studies revealed that overexpression of ACD2 in citrus promotes CLas multiplication (Pang et al., 2020). However, ACD2-edited or ACD2-silenced citrus roots have never been obtained using A. tumefaciens-mediated genetic transformation because ACD2 family genes are involved in the regulation of cell death and all corresponding transgenic shoots ultimately die. In the present study, we successfully obtained ACD2-edited citrus roots using K599-mediated genetic transformation ( Figures 5A–C ), indicating that this technology is advantageous for investigation of specific genes to obtain “hard-to-get” transgenic root tissue. In addition, A. rhizogenes-mediated citrus genetic transformation facilitates evaluation of tissue-specific promoters. When identifying exogenous genes that enhance resistance against a phloem-limited pathogen, a phloem-specific promoter is critical to reduce the impact of these genes on citrus biological traits (Dai et al., 2004; Tzean et al., 2020). In Arabidopsis, the companion cell-specific AtSUC2 promoter has been widely used in gene function analysis of phloem-related genes (Paultre et al., 2016). In the present study, the homologous sequence of the AtSUC2 promoter was identified in citrus and designated CsSUC2pro. Subsequently, we constructed three binary plasmids containing the GFP and GUS genes, of which CsSUC2pro, 35S, or no promoter was inserted at the 5′ end of GUS ( Figure 5D ). Using K599-mediated genetic transformation, we obtained transgenic citrus hairy roots harboring these three plasmids ( Figure 5E ). The results of GUS staining showed that all cells of 35S-GUS transgenic roots were stained blue, whereas only cells located in the phloem of CsSUC2-GUS transgenic roots were stained blue, and no cells of GUS (no promoter) transgenic roots were stained ( Figure 5F ).
Virus-induced gene silencing (VIGS), virus-mediated genome editing, or foreign gene overexpression technology have been widely used for gene function analysis in plants in recent years (Velázquez et al., 2016; Ellison et al., 2020). However, a viral vector cannot be transiently expressed in citrus leaves via Agrobacterium infiltration. Application of this technology to citrus is usually reliant on the use of tobacco (Nicotiana spp.) for virion enrichment. Given that K599-mediated genetic transformation of citrus is highly efficient, the feasibility of K599-mediated viral inoculation in citrus was examined. First, the Citrus leaf blotch virus (CLBV)-ChlI vector containing a partial sequence of ChlI (Magnesium chelatase subunit I) was constructed for ChlI silencing in citrus ( Figure 6A ). Citrus medica explants were infiltrated with K599 harboring the CLBV-ChlI viral vector. Agroinfiltrated explants displayed photobleaching phenotypes in new leaves after 3 months ( Figure 6B ). RT-qPCR results showed that the expression level of ChlI in CLBV-ChlI-transfected leaves was significantly downregulated, indicating that successful K599-mediated viral inoculation could be used for VIGS ( Figure 6C ). Subsequently, the CLBV-FT vector ( Figure 6D ) containing the full-length sequence of the Arabidopsis thaliana FLOWERING LOCUS (FT) gene was constructed for FT overexpression in citrus based on a previously published method (Velázquez et al., 2016). The CLBV-FT vector was agroinfiltrated using 2-month-old C. medica seedlings, and the citrus infected with CLBV-FT flowered 9 months later ( Figure 6E ). The RT-qPCR results showed that the expression level of FT in CLBV-FT-transfected C. medica was significantly upregulated ( Figure 6F ).
Currently, A. tumefaciens, electroporation, particle bombardment, and RNA interference are used for citrus genetic transformation, but these methods are laborious, expensive, time-consuming, and inefficient. The limitation of these methods is largely due to their reliance on tissue culture, which requires an aseptic environment. Using A. tumefaciens-mediated genetic transformation as an example, the acquisition of transgenic citrus requires 4- to 5-week-old seedlings germinated on Murashige and Skoog medium, co-incubation of explants with A. tumefaciens, and shoot regeneration, and each step involves tissue culture under aseptic conditions (Orbović and Grosser, 2015). However, only few researchers have studied the genetic transformation of citrus bypassing tissue culture. With regard to A. tumefaciens, Chun-zhen Cheng developed an in planta genetic transformation approach for pomelo (Citrus maxima) and obtained transgenic plants using this method 3 months post-transformation (Zhang et al., 2017). On the other hand, virus-mediated genome editing in tobacco, Arabidopsis, and wheat has been reported (Ellison et al., 2020; Ghoshal et al., 2020; Ma et al., 2020; Luo et al., 2021). However, the aforementioned in planta genetic transformation technology requires 3–5 months to obtain transformants. In addition, virus-mediated genome-editing technology for woody plants has not been reported to date. In recent years, the technology for obtaining transgenic roots using A. rhizogenes has been well developed in many herbaceous plants, such as soybean (Kereszt et al., 2007; Cao et al., 2009; Cheng et al., 2021), grain amaranth (Castellanos-Arévalo et al., 2020), rice (Raineri et al., 1990), and maize (Ishida et al., 2007). However, no report of A. rhizogenes-mediated endogenous gene editing bypassing tissue culture in citrus is available. In the present study, the genetic transformation procedure for citrus using A. rhizogenes bypassing tissue culture required only 2–4 weeks (C. medica) or 1–2 months (C. limon and citrange ‘Carrizo’). The explants used in this study are branches, which are highly convenient to obtain. Subsequently, we established a highly efficient, convenient, and cost-effective genome-editing technology system in citrus using this technology. In addition, the procedure can be used for efficient inoculation with viral vectors. Furthermore, hairy roots induced by A. rhizogenes often develop from single cells, which leads to a lower incidence of chimerism in transgenic roots (Roychowdhury et al., 2017). This phenomenon was confirmed using the technology in the present study. In our study, the transcriptome analysis of citrus callus induced by A. rhizogenes revealed that hormone (IAA) pathway was significantly triggered, which provide evidence to explain the highly efficient transformation rate in citrus. To improve the efficiency of verification of transgenic hairy roots, the binary vector (1380-Cas9-HA) used contains the GFP reporter gene. Thus, the hairy roots can be identified by hand-held excitation light detection of GFP, which is time- and labor-saving. The establishment of a highly efficient genetic transformation technology mediated by A. rhizogenes for multiple citrus species is of great importance for gene functional analysis in citrus. In summary, we established a highly efficient genetic transformation technology bypassing tissue culture for citrus, which can be used for genome editing, gene overexpression, and virus-mediated gene function analysis. The advantages of this technology are as follows: (1) the explant used for transformation are citrus branches, which is convenient to obtain; (2) the transformation process does not involve tissue culture and thus is convenient to implement; (3) the process is time-saving (2–8 weeks); (4) the procedure is less labor demanding (as few branches are required); (5) a high frequency of positive transformants is obtained (~57%, C. medica); (6) gene transformation or genome editing are achieved with high efficiency ( Table S10 ). The problems that may be encountered during the experiment and the corresponding solutions were listed in Table S11 . We anticipate that by removing the high cost, heavy workload, long experimental period, and other technical obstacles, this genetic transformation technology will be a valuable tool for routine investigation of endogenous and exogenous genes in citrus.
The transcriptome data presented in the study are deposited in the NCBI repository, accession number PRJNA800116.
Conceptualization, HM, SD, XS; writing—original draft preparation, HM; writing—review and editing, SD, XS, FG, NW; supervision, HM, SD, XS, JL; project administration, HM, XM; transcriptome analysis, YG; experiment, XM, MW, NL, SH, ML; plants maintenance, XM, HX, KX, MW; funding acquisition, JL, ML. All authors contributed to the article and approved the submitted version.
This work was supported by the National Natural Science Foundation of China (32202427 and 32002021); the Key Project for New Agricultural Cultivar Breeding in Zhejiang Province, China (2021C02066-1); The Major Science and Technology R& D Program of Jiangxi Province (20194ABC28007).
We thank Honghong Deng, Fabieli Irizarry, Ming Huang, Chen Ling, Jiaying Fang, Hujing Wang, and Qibin Yu for assistance with plant materials and laboratory activities.
JL was employed by Natural Medicine Institute of Zhejiang YangShengTang Co., LTD. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647161 | Yubo Wang,Minghui Zhang,Kuiling Dong,Xiaojuan Yin,Chunhui Hao,Wenge Zhang,Muhammad Irfan,Lijing Chen,Yong Wang | Metabolomic and transcriptomic exploration of the uric acid-reducing flavonoids biosynthetic pathways in the fruit of Actinidia arguta Sieb. Zucc. | 27-10-2022 | flavonoids,Actinidia arguta Sieb.Zucc.,transcriptome,widely targeted metabonomics,biosynthetic pathway | Flavonoids from Actinidia arguta Sieb. Zucc. can reduce uric acid in mice. However, the molecular basis of its biosynthesis is still unclear. In this paper, we used a combination of extensively targeted metabolomics and transcriptomics analysis to determine the types and differences of flavonoids in the fruit ripening period (August to September) of two main cultivated varieties in northern China. The ethanol extract was prepared, and the potential flavonoids of Chrysin (Flavone1), Rutin (Flavone2), and Daidzein (Flavone3) in Actinidia arguta Sieb. Zucc. were separated and purified by HPD600 macroporous adsorption resin and preparative liquid chromatography. The structure was identified by MS-HPLC, and the serum uric acid index of male Kunming mice was determined by an animal model test.125 flavonoids and 50 differentially regulated genes were identified. The contents of UA (uric acid), BUN (urea nitrogen), Cr (creatinine), and GAPDH in mouse serum and mouse liver glycogen decreased or increased in varying degrees. This paper reveals the biosynthetic pathway of uric acid-reducing flavonoids in the fruit of Actinidia arguta Sieb. Zucc., a major cultivar in northern China, provides valuable information for the development of food and drug homologous functional foods. | Metabolomic and transcriptomic exploration of the uric acid-reducing flavonoids biosynthetic pathways in the fruit of Actinidia arguta Sieb. Zucc.
Flavonoids from Actinidia arguta Sieb. Zucc. can reduce uric acid in mice. However, the molecular basis of its biosynthesis is still unclear. In this paper, we used a combination of extensively targeted metabolomics and transcriptomics analysis to determine the types and differences of flavonoids in the fruit ripening period (August to September) of two main cultivated varieties in northern China. The ethanol extract was prepared, and the potential flavonoids of Chrysin (Flavone1), Rutin (Flavone2), and Daidzein (Flavone3) in Actinidia arguta Sieb. Zucc. were separated and purified by HPD600 macroporous adsorption resin and preparative liquid chromatography. The structure was identified by MS-HPLC, and the serum uric acid index of male Kunming mice was determined by an animal model test.125 flavonoids and 50 differentially regulated genes were identified. The contents of UA (uric acid), BUN (urea nitrogen), Cr (creatinine), and GAPDH in mouse serum and mouse liver glycogen decreased or increased in varying degrees. This paper reveals the biosynthetic pathway of uric acid-reducing flavonoids in the fruit of Actinidia arguta Sieb. Zucc., a major cultivar in northern China, provides valuable information for the development of food and drug homologous functional foods.
Flavonoids are one of the main secondary metabolites in Actinidia arguta Sieb. Zucc. Structurally, they are mainlyflavonols, dihydroflavonols, 3-o-flavonoid glycosides, and their derivatives. Wojdyło and Nowicka (2019) identified the polyphenol compounds in Actinidia arguta Sieb. Zucc. and obtained 16 flavonols, 7 flavonols, 7 phenolic acids, and 1 anthocyanin. To explore the relationship between metabolite changes and fruit color changes. Li et al. (2018b) carried out transcriptomics and metabolomics analysis on the flesh of two kinds of Actinidia arguta Sieb. Zucc. and identified a variety of flavonoids such as bitter bracteachin, luteolin, dihydromyricetin, anthocyanin, geranium, delphinidin, and (-) - epigallocatechin. Jang et al. (2009) isolated two new flavonoids with γ-lactams from the roots of Actinidia arguta Sieb. Zucc., which are flavan-3,6-(2-pyrrolidinome-5-yl)-(−)-epicatechin and 8-(2-phrrolidinone-5-yl)-(−)-epicatechin and also get proanthocyanidin B-4.Flavonoids have physiological functions such as antioxidant, antiviral, prevention and treatment of cardiovascular and cerebrovascular diseases, prevention of hyperuricemia, liver protection, and immunity (Latocha et al., 2013; Hu et al., 2016; Jiang et al., 2020). Hyperuricemia (HUA) is a pathological state in which UA levels in the blood increase continuously or the blood is supersaturated with UA. The number of patients with HUA in China exceeded 17 million in 2017, and the data shows a rapid increase in cases, with an annual growth rate of 9.7%. Gender, age, race, and lifestyle habits all affect the incidence of HUA (Lanaspa et al., 2011; Chen et al., 2022; Maruhashi et al., 2022; Tsai et al., 2022; Yu et al., 2022). The key cause of primary HUA is a combination of low UA excretion and high UA production. A majority (67%) of UA in the human body is produced by the catabolism of nuclear proteins, nucleic acids, and other substances in the body; the remaining 33% comes from purines in food (Lai et al., 2021; Lee et al., 2022; Mccormick et al., 2022). Adenosine deaminase (ADA) and xanthine oxidase (XO) are the key enzymes that regulate the production of UA during the catabolism of purine substances to UA. ADA is a sulfhydryl enzyme that catalyzes the reaction of adenine nucleosides to produce hypoxanthine nucleosides. Hypoxanthine is then produced through the action of nucleoside phosphorylase, and hypoxanthine is finally oxidized by the flavin protease XO to produce UA and XO is a flavin protease (Zhang et al., 2018; Jiang et al., 2020; Le et al., 2020; Michael et al., 2020; Xu et al., 2021). In previous reports, ten flavonoids belonging to quercetin, isorhamnetin, and kaempferol were detected in the leaves of Changjiang No. 1 (CJ-1) by transcriptomics and metabolomics methods (Tan et al., 2021). Metabolomics and transcriptomics analyses provide us an opportunity of comprehending the flavonoid biosynthesis of the developing seed of Tartary Buckwheat. A total of 234 flavonoids were identified, containing 10 isoflavones, of which 80 flavonoids accumulated prominently in the period of seed development (Li et al., 2019a). The flavonoid biosynthesis of different colored flowers in safflower was analyzed by metabonomics and transcriptomics. Metabolic analysis showed that there are great differences in flavonoid metabolites among different colored safflower (Wang et al., 2021) Natural bioflavonoids have a small molecular weight and can penetrate adipose tissue, pass through the blood-brain barrier, and are quickly absorbed by the human body these are the material basis for bioflavonoids to play a pharmacological role (Zuo et al., 2012; Yu et al., 2015). In recent years, scholars both domestically and internationally have carried out research into the active components and functional mechanism of such botanical drugs and found that flavonoids such as morin, quercetin, luteolin, and kaempferol have significant effects in treating hyperuricemia (HUA) and gout (Mo et al., 2007; Ouyang et al., 2021). Studies have shown that flavonoids can prevent HUA by inhibiting the activity of xanthine oxidase (XO) and by promoting the excretion of uric acid. Gouty arthritis can be prevented by inhibiting the release of inflammatory transmitters by neutrophils and by inhibiting the expression and secretion of inflammatory cytokines, which are induced by urate crystallization (Martin et al., 2010; Li et al., 2022a). Making full use of Actinidia arguta Sieb. Zucc. flavonoids to develop flavonoid products has broad application prospects in the field of medicine and food homology. At present, the uric acid-reducing activities and biosynthetic pathways of its flavonoids, aspen, rutin, and daidzein, have not been systematically analyzed. Some scholars analyzed the metabolomes and transcriptomics of the flesh of two kinds of Actinidia arguta Sieb. Zucc. at different fruit development stages, namely “Hong Bao Shi Xing” and “Yong Feng No. 1”. The results showed that AaF3H, AaLDOX, AaUFGT, AaMYB, AabHLH and AaHB2 were the most likely candidate genes to regulate the biosynthesis of flavonoids. Meanwhile, in another study, it was found that the AaLDOX gene may be the key gene controlling anthocyanin biosynthesis in the flesh of “Tian Yuan Hong” Actinidia arguta Sieb. Zucc., which promotes anthocyanin accumulation and eventually leads to red flesh (Li et al., 2018a; Li et al., 2018b). They then screened miR858 involved in anthocyanin biosynthesis through high-throughput sequencing of microRNA and proved miR858 was a negative regulator of anthocyanin biosynthesis by inhibiting the target gene AaMYBC1 in red Actinidia arguta Sieb. Zucc. (Li et al., 2019b; Li et al., 2020).Studies have shown that two interacting transcription factors AcMYB123 and AcbHLH42 and another AcMYB10 have a regulatory effect on the biosynthesis of tissue-specific anthocyanins in the endocarp of Actinidia arguta Sieb. Zucc. (Wang et al., 2019b; Yu et al., 2019).Yanfei Liu et al. showed the cMYBF110-AcbHLH1-AcWDR1 complex directly targeted the promoter of the anthocyanin synthesis gene and promoted the activities of AcMYBF110, AcbHLH1, and AcWDR1. The AcMYBF110-AcbHLH4/5-AcWDR1 complex amplified the regulatory signal of the first MBW complex by activating the promoter of AcbHLH1 and AcWDR1 and indirectly participated in the regulation of anthocyanin synthesis (Liu et al., 2021). In this study, the uric acid-lowering effect of flavonoid extract from Actinidia arguta Sieb.Zucc. was evaluated in vitro. Furthermore, the types, quantities, and differences between flavonoids in the fruits of two important varieties of Actinidia arguta Sieb.Zucc. cultivated in Northern China were determined through a combination of extensive targeted metabolomics and transcriptomics analyses. The biosynthetic pathways and structural genes involved in regulating the flavonoid compounds Chrysin (Flavone1), Rutin (Flavone2), and Daidzein (Flavone3), which have uric acid-reducing activity, were analyzed and identified. This provides valuable information for further improving the fruit quality of Actinidia arguta Sieb.Zucc., breeding new varieties, and developing food and drug homologous functional foods from Actinidia arguta Sieb.Zucc.
8-year-old Actinidia arguta Sieb.Zucc. the mature fruit of Qssg and Lc varieties was obtained from North China, which mature from August to September. Fresh fruit was quickly frozen for the next experiment. 95% ethanol, petroleum ether, n-butanol, rutin standard, HPD600 macroporous adsorption resin, and absolute ethanol are all analytical pure. Allopurinol sustained release capsule, purchased from Heilongjiang aolidanede Pharmaceutical Co., Ltd; Ethambutol hydrochloride tablets, purchased from Hangzhou Minsheng Pharmaceutical Co., Ltd; Adenine, purchased from American sigma company; Acetaminophen sustained release tablets, purchased from Shanghai Johnson & Johnson Pharmaceutical Co., Ltd; UA (uric acid) kit, bun (urea nitrogen) kit, Cr (creatinine) kit, GAPDH (glyceraldehyde 3-phosphate dehydrogenase) kit and glycogen kit were purchased from Quanzhou konodi Biotechnology Co., Ltd. Kunming white mice, weighing 19-21g, were purchased from Shenyang Changsheng Biology Co., Ltd.Before starting the experiment, they were settled in the laboratory environment for seven days. 6 animals for one cage (320× 180 × 160 cm), according to the 12-hour/12-hour light and dark schedule. Temperature: 22 ± 2 °C; Relative humidity: 55 ± 5% and food and water were given in the standard. Our experiments were conducted based on the requirements of the institutional animal care committee of Nanjing University and the China Animal Care Council of Nanjing University [SYSK (SU) 2009 – 0017].
A UPLC-MS/MS analysis conducted by Metware Biotechnology Co., Ltd. (Wuhan, China) detected 786 metabolites. To prepare the biological samples for analysis, they were first freeze-dried in a vacuum freeze dryer (Scientz-100F). Next, the samples were ground (30 Hz, 1.5 minutes) to powder. Then 100 mg of the resulting powder was dissolved in 1.2 ml of 70% methanol extract and vortexed 6 times, once every 30 minutes for 30 seconds, and placed into a 4°C refrigerator overnight. Samples next underwent centrifugation (rotating speed 12000 rpm, 10 minutes) followed by absorption of the resulting supernatant. Finally, the samples were filtered through a 0.22 μM microporous membrane and stored in an injection bottle for UPLC-MS/MS analysis. Using a self-built MWDB (metal database), a qualitative analysis of substances was carried out based on the secondary spectrum information using triple quadrupole multiple reaction monitoring (MRM) mass spectrometry (Fraga et al., 2010). Software Analyst 1.6.3 was used to process mass spectrometry data. We employed principal component analysis to preliminarily explore the general metabolic differences and variabilities between samples. The PCA results display a trend of metabolomics separation between groups, suggesting metabolomics differences between sample groups (Chen et al., 2009). The metabolomics data were analyzed according to the OPLS-DA model, and score maps were drawn to further show the differences between each group (∣log2 (fold change) ∣ ≥ 1) (Thévenot et al., 2015). Metabolites in each sample were analyzed, with three independent biological replicates.
Take the frozen Actinidia arguta Sieb.Zucc. and wash it with distilled water after melting. Dry it in the air, slice it, and grind it in a homogenizer until it is homogenized. After ethanol extraction, centrifugation, filtration, and concentration, the crude extract of total flavonoids of Actinidia arguta Sieb.Zucc. was obtained, which was used for standby. A Rutin standard curve was generated to determine the concentration of flavonoids in the crude extract. Macroporous resin and preparative liquid chromatography were then used for separation and purification. In this experiment, flavonoids were identified by electrospray spray mass spectrometry. Thirty-six male Kunming mice were separated into six groups at random after seven-day adaptive feeding in the laboratory environment at 22 ± 2°C and a 55 ± 5% relative humidity. The groups were as follows: blank control group, model control group, positive control group (Allopurinol), Chrysin(Flavone1), Rutin (Flavone2), and Daidzein (Flavone3). The blank group was gavaged with distilled water, and the animals in all other groups were gavaged with a 2.5% PAPA suspension for seven consecutive days. Then the mouse model of hyperuricemia was induced by gavage with 100 mg/kg adenine and 250 mg/kg ethambutol. Animals in the treatment groups (i.e., groups other than the blank and model controls) were given the same volume of distilled water, and the appropriate drugs were administered by gavage. The dose of flavonoids was 550 mg/kg. The dosage of allopurinol tablets was 33.3 mg/kg (administration volume: 1 ml/100 g) for 23 consecutive days. One hour after treatment administration on the 7th and 15th days, eyeball blood was collected. The serum was centrifuged and serum levels of uric acid (UA), urea nitrogen (BUN), creatinine (Cr), glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and hepatic glycogen were measured using kits.
mRNA with PolyA tail was enriched using Oligo (dT) magnetic beads, and then chemically fragmented. Using the resulting short segment RNA as a template, the first strand of cDNA was synthesized with six base random primers/hexamers. Subsequently, the double-stranded cDNA was synthesized by adding buffer, dNTPs, and DNA polymerase I, and purified with ampure XP beads, and then was subjected to end repair, A-tail addition, and connect sequencing connector, and fragment size was selected with ampure XP beads. PCR enrichment yielded the final cDNA library. Qubit2.0 was used for preliminary quantification, Agilent 2100 was used to detect the insert size of the library, and the Q-PCR method quantified the effective concentration of the library (> 2nm). After passing the library inspection, the libraries were pooled according to the target offline data volume, and Biomarker Technology Co., Ltd. (Beijing, China) conducted the sequencing using the Illumina novaseq platform.
Clean reads for subsequent analysis were obtained following raw data filtering, sequencing error rate inspection, and GC content distribution inspection. The clean reads were spliced with Trinity (Grabherr et al., 2011), and stored in FASTA format. Unigene corset (Davidson and Oshlack, 2014) hierarchical clustering was used to obtain the longest cluster sequence, which was then compared with the KEGG, NR, Swiss-Prot, GO, COG/KOG, and Trembl databases using DIAMOND (Buchfink et al., 2015) BLASTX software. After predicting the amino acid sequence, HMMER software was used to compare the sequence with the Pfam database to obtain the Unigene annotation information. RSEM (Li and Dewey, 2011) software and bowtie2 (Langmead and Salzberg, 2012) were used to compare the statistical results. FPKM (fragments per kilobase of transcription per million fragments mapped) was taken as an index to detect the level of transcripts or gene expression. DESeq2 (Love et al., 2014; Varet et al., 2015) was used to obtain the differential expression between the two biological conditions. The FDR (false discovery rate) value was <0.05 and ∣ log2 (folding change) ∣ ≥ 1 was used as the threshold of significant expression difference. Through GOannotation and KEGG pathway analysis, the identified DEG was further enriched and analyzed.
To validate the RNA-Seq data and examine the expression of flavonoid biosynthesis-related genes, qRT-PCR was carried out as described in the previous literature. The amplification cycle procedure was as follows: the reverse transcription operation was carried out using an Aidlab company’s kit (TUREscript 1st Stand cDNA SYNTHESIS Kit), and the 20ul reaction system was adopted. The reverse transcription reaction conditions were 42 °C for 40min and 65 °C for 10min, and the fluorescent quantitative PCR procedure was 95 °C for 3min, 95 °C for 10s, and 60 °C for the 30s. The relative gene expression of each sample and group was calculated by using 2-△△Ct with actin as the internal reference. Each sample was performed in triplicates. Table S1 lists the primers used in qRT-PCR.
The correlation coefficient was calculated for the content of flavonoids and the transcriptional changes of both differentially expressed flavonoids and differentially expressed genes. Both are rich in biosynthesis pathways of flavonoids, the flavonol (ko00941), flavonoid (ko00942), and secondary metabolite (ko00943). Cytoscape2.8 was used to visualize the interaction network between DEGs and differentially accumulated flavonoids to identify the structural genes involved in the regulation of flavonoids, such as Chrysin (Flavone1), Rutin (Flavone2), and Daidzein (Flavone3), with uric acid reducing activity.
A total of 125 flavonoids were identified by qualitative and quantitative analysis of the Qssg and Lc metabolite spectrum. The mature fruits of Qssg and Lc are shown in Figure 1 ( Table S2 ). PCA revealed these two varieties to be significantly different; 68.61% of the differences between the samples could be explained by PC1 (40.7%) and PC2 (27.91%), suggesting their pattern change of flavonoid accumulation ( Figure 2A ). Hierarchical cluster analysis (HCA) further confirmed the difference between the two main samples ( Figure 2B ). Chrysin (Flavone1), Rutin (Flavone2), and Daidzein (Flavone3) in both Qssg and Lc fruits were isolated and purified. A male Kunming mouse animal model was used to carry out a uric acid lowering activity test ( Figures 2C–G ). Compared with the model control group, the three selected flavonoids significantly reduced UA after one week (p < 0.01) ( Figure 2C ). At two weeks, Daidzein significantly reduced UA (p < 0.05), but there was no significant difference in UA activity between Chrysin and Rutin. Daidzein had relatively stable biological activity in reducing UA. Compared with the blank control group, BUN was not significantly increased in the model control group ( Figure 2D ). Compared with the model control group, only Daidzein significantly reduced BUN (p < 0.05), and there was no significant difference in the BUN-reducing activity of the other two flavonoids. Thus, Daidzein had a higher biological capacity to reduce BUN than Chrysin or Rutin did. Daidzein also significantly reduced Cr compared to the model control group (p < 0.05), but there was no significant difference in Cr reducing activity between the other two Flavone treatment groups ( Figure 2E ). Compared with the model control group, Rutin, Daidzein, and the positive control group (treated with allopurinol) all had significantly lower levels of GAPDH activity (p < 0.01), meaning that only the Chrysin treatment group showed no significant difference ( Figure 2F ). Daidzein also significantly increased liver glycogen compared to the model control group (p < 0.01); in contrast, Rutin significantly decreased liver glycogen (p < 0.01), and there was no significant difference in liver glycogen in those treated with Chrysin ( Figure 2G ).
Orthogonal signal correction (OSC) and PLS-DA techniques were used to identify variable differences. Abundances between samples of the 125 flavonoids were compared using an OPLS-DA model. The model was able to discriminate between the two varieties as both samples fell outside the confidence interval, with the Lc samples to the left and the Qssg samples to the right of the interval. The OPLS-DA yielded two principal components with contribution rates of 62.1% and 9.74% (R2x = 0.807, R2y = 0.998 [p = 0.29], Q2 = 0.869 [p < 0.005]). This result was verified by 200 replicate analyses. The differential flavonoids were screened according to VIP analysis ( Figures 3A, B ). The pathway enrichment analysis of flavonoids was carried out through KEGG (Kyoto Encyclopedia of genes and genes) database. The results showed that the 125 identified flavonoids were mainly distributed in three metabolic pathways,(1) flavonoid and flavonol biosynthesis pathways, primarily for kaempferol-3-o-neohesperidin Luteolin-7-o-neohesperidin, quercetin-3-o-sangbu diglycoside, quercetin-3-o-(2”-o-xylosyl)-rutoside, and kaempferol-3-o-rutoside;(2) flavonoid biosynthesis pathways, mainly including naringin-7-o-glucoside, isolyceride, and chrysin;(3) secondary metabolite biosynthesis pathways, primarily kaempferol-3-o-rutoside and daidzein ( Figure 3C ). In a mass spectrum graph, the mass charge ratio (M/z) of the ion increases from left to right, and the abscissa of an ion with a single charge is the mass of its ion; the ordinate represents the intensity of the ion current, usually expressed in terms of relative intensity. The elution rate is 68.8% for 50% ethanol, 51.2% for 60% ethanol, and 48.2% for 70% ethanol. Based on our test data, the peak value of flavonoids eluted by 50% ethanol was more and the elution rate was higher. Therefore, only positive ion mode mass spectra of flavonoids eluted by 50% ethanol are discussed here. The mass spectrum information for full scanning mode was m/Z 609, 301, and 175, the multi-reaction detection mode was m/Z [609/301], and the neutral loss was 308u. By comparing the retention times, multi-stage mass spectrum fragments, excimer ion peaks, and other information with reference substances, we determined that the molecular ion peak was m/z 609 for Rutin, m/z 254 for Chrysin, and m/z 254 for soybean isoflavone ( Figures 3D–F ).
Nine cDNA libraries were generated for high-throughput RNA-Seq analysis to further examine the potential molecular mechanisms of flavonoid biosynthesis in Actinidia arguta Sieb.Zucc. Each library obtained 281,758,296 clean reads from 439,776,908 to 43,580,384 and 269,732,354 from 41,844,326 to 41,708,692 ( Table S3 ). The Q30 percentage (including sequences having an error rate < 0.1%) for each library exceeded 91%, with 47.47% GC content on average. Among the clean reads, between 78.54% and 80.76% could be mapped to the reference genome. A total of 43,686 unique genes wereidentified, supplying high-quality RNA-Seq data for further analyses
To identify the DEGs in the mature fruits of two kinds of Actinidia arguta Sieb.Zucc., the correlation coefficient between gene expression profile clustering and biological duplication was first analyzed ( Figure 4A ), indicating there was a large number of differential expressions between different samples. The gene expression correlation coefficient level between biological replication of all samples were greater than 0.8, indicating that biological replication is very good, and the data can be further used to determine DEG. According to the FDR (false discovery rate) values <0.05 and ∣ log2 (fold change) ∣ ≥ 1 between the two sample groups as the threshold of significantly different gene expression, it was determined that 6497 was up-regulated and 5153 was downregulated between the two sample groups ( Figure 4B ). The most abundant items among the 25 biological process categories were metabolic processes, cellular processes, and single biological processes. The most representative terms among the 13 cell component categories were cell part, cell, and organelle. Among the 10 molecular functional categories, the most common terms were catalytic activity, binding, and transporter activity ( Figure 5A and Table S4 ). 4024 DEGs were allocated to 142 KEGG paths ( Table S5 ). The flavonoid (ko00941), flavonol (ko00944), and secondary metabolite (ko01110) biosynthesis pathways were all enriched, with the flavonoid and secondary metabolite pathways being significantly enriched. Flavonoid biosynthesis and secondary metabolite biosynthesis are present in the significant enrichment pathway ( Figure 5B and Table S6 ). The enriched pathways could be further divided into five categories: cellular processes, genetic information processing, environmental information, metabolism, and tissue systems. The metabolism category contained the most pathways and the highest number of DEGs were involved in amino acid biosynthesis (ko01230; 185 genes), carbon metabolism (ko01200; 190 genes), and sucrose and starch metabolism (ko00500; 207 genes).
KEGG analysis and gene functional annotations were two strategies to determine which DEGs encode enzymes related to flavonoid biosynthesis, flavonoid and flavonol biosynthesis, and secondary metabolite biosynthesis. The results showed that 50 DEG genes, including 6 UGT9491 genes, 16 LOC genes, 2 AT2 genes, 2 CHS, 3 C4Ha genes, 1 HCT gene, 1 CCoAOMT gene, 1 F3H gene, 2 LAR2 genes, 2 4CL, 1 VIT, 4 PAL, and 1 GSCOC gene, were significantly up-regulated in the two varieties, 2 CFOL genes, 1 CHIa gene, 1 C4Ha gene, 1 DFR gene, 1 LAR2 gene,1 LOC gene, and 1 LSAT gene were significantly down-regulated in the two varieties ( Table S7 ).
To test the expression of DEGs related to flavonoid biosynthesis in the fruits of the two varieties at maturity, Twenty structural genes (1 CsUGT134, 1 LAR2, 2 C4H, 1 CFOL, 1 CHI, 2 LOC, 1 GSCOC, 1 CCoAOMT, 1 DFR, 1 AT2,2 4CL,1 VIT,4 PAL,1 CHS) were quantified by qRT-PCR. The detected genes were highly consistent between the RNA-Seq and qPCR results according to Reverse transcriptase polymerase chain reaction results( Figure 6 ). The relative expression of CFOL, CsUGT134, PAL, C4Ha, CHS, and LOC in Lc was significantly higher than that in Qssg; the relative expression of CHIa, LAR2, GSCOC, and CCoAOMT in regulating flavone content in Qssg varieties was significantly higher than that in Lc varieties. The result confirmed the transcriptomics-derived data.
To comprehend the pathway and regulatory structural genes of flavonoid biosynthesis in the two main varieties of Actinidia arguta Sieb.Zucc. fruit, the quantitative changes of flavonoids and transcripts in the fruit ripening stage of Actinidia arguta Sieb.Zucc. were tested and analyzed by studying the interaction between transcriptomics and metabolomics. Based on the results of DEGs and Dems rich in flavonoid biosynthesis pathway, it was annotated that 20 structural genes and regulatory groups show a higher correlation with the biosynthesis pathway of flavonoid compounds Chrysin, Rutin, and Daidzein ( Table S8 ). Their interaction network is shown in ( Figures 7A–C ). According to the analysis of the pathway diagram in Figure 7A , DFR, F3’H, and FLS compete with the substrate dihydrokaempferol DHK. According to the gene expression analysis in Figure 7C , FLS has relatively high expression and high activity of its coding enzyme, which boosts the synthesis of Rutin. It was also found that the relative content of Rutin accumulated in the sample was high ( Figure 7B ). The LOC expression was comparatively low, and the accumulated Chrysin and Daidzein substances were relatively low. According to the analysis in the pathway diagram in Figure 7A , DFR, F3’H, and FLS compete with the substrate dihydrokaempferol DHK. According to the gene expression analysis in Figure 7C , FLS has relatively high expression and high activity of its coding enzyme, which finally promotes the synthesis of Rutin. The relative content of Rutin accumulated in the sample was high ( Figure 7B ); the expression of LOC and the relative contents of Chrysin and Daidzein was relatively low.
Flavonoids are substances that have been studied in depth and at length in traditional Chinese medicine. They exist widely throughout the kingdom Plantae. Many studies have focused on the abilities of flavonoids to reduce blood lipid levels, inhibit lipid peroxidation, relieve coughs, eliminate phlegm and asthma, and exert anti-tumor, anti-hepatotoxicity, anti-inflammatory, analgesic, antibacterial, and antispasmodic effects. Substances known to reduce uric acid include quercetin, luteolin, apigenin, puerarin, catechins, and dyestuffs, but there have been few studies into the uric acid-lowering activities of Rutin, Chrysin, and Daidzein. In this study, extensive targeted metabolomics methods were used to compare the flavonoids present in the fruit of two varieties of Actinidia arguta Sieb.Zucc. cultivated in Northern China. 9 classes and 125 kinds of flavonoids were detected in the fruits of the two varieties; these included 39 kinds of differentially accumulated flavonoids, accounting for 31.2% of the total flavonoids detected. This demonstrated that there were significant differences in flavonoid composition and accumulation between different varieties. The types and proportions of different flavonoids found in the fruits were as follows: 51.2% flavonols, 14.4% flavonoids, 9.6% dihydroflavonoids, 9.6% flavonols, 1.6% chalcones, 2.4% flavone carboglycosides, 4.8% dihydroflovonols, 0.8% isoflavones, and 5.6% procyanidins. They are mainly distributed in 3 metabolic pathways, among which the flavonoid and flavonol biosynthesis pathways mainly include kaempferol-3-O-neohesperidin, luteolin-7-O-neohesperidin, quercetin- 3-O-Sambubiglycoside, quercetin-3-O-(2”-O-xylosyl)rutinoside, kaempferol-3-O-rutinoside; quercetin-3-O-rutinoside (Rutin); the flavonoid biosynthesis pathway mainly includes naringenin-7-O-glucoside, isoflavin, and Chrysin; the secondary metabolite biosynthesis pathway mainly includes kaempferol-3-O-rutinoside, Daidzein. The relative content of flavonoids in the fruit of the variety Lc was increased by comparing it with the fruit of the variety Qssg Actinidia arguta Sieb. Zucc.These data provide a strong basis for the enrichment study of flavonoids in Actinidia arguta Sieb. Zucc. and also provide a valuable theoretical and practical basis for breeding new varieties of Actinidia arguta Sieb.Zucc. and developing food and drug homologous functional foods. The flavonoids Rutin, Chrysin, and Daidzein were isolated from the fruit of the main variety of Actinidia arguta Sieb. Zucc. cultivated in northern China and tested for their uric acid-reducing activity. The results showed that Chrysin, Rutin, and Daidzein could reduce serum levels of UA, BUN, Cr, and GAPDH in mice to varying degrees. We, therefore, speculate that uric acid production in mice (in response to purine-rich food ingestion or catabolism of substances such as nuclear proteins and nucleic acids) may be inhibited by reducing the activities of ADA and/or XO. Increases in liver glycogen content in mice may be due to inhibition of glycolysis by flavonoids through the promotion of gluconeogenesis, reduction of free glucose in the blood, promotion of the remedial synthesis pathway of nucleotides, and decreases in serum uric acid content. This study, therefore, has strong practical significance for the prevention and control of human HUA through the comprehensive development of Actinidia arguta Sieb.Zucc. as a treatment. There are two general kinds of genes involved in the biosynthesis of plant flavonoids: structural genes, which encode enzymes that catalyze the flavonoid biosynthesis, and regulatory genes, which regulate the structural genes’ expression levels (Gupta et al., 2011; Li et al., 2012; Chen et al., 2017). Transcriptomics analysis identified 43,686 genes involved in flavonoid biosynthesis in the fruits of two Actinidia arguta Sieb. Zucc. varieties, including 11,650 differentially expressed genes. These genes regulate metabolism, cellular processes, genetic information processing, environmental information, and tissue systems in the Actinidia arguta Sieb. Zucc. fruit. KEGG enrichment analysis and gene function annotation identified 20 structural genes encoding 50 known flavonoid biosynthesis-related enzymes, with most of the genes highly correlated between the qPCR and RNA-seq datasets. The correlation analysis between the transcriptomics and metabolic spectrum revealed the expression level of some structural genes to be bound up with the accumulation of particular flavonoids, indicating that the expression of these flavonoid biosynthesis genes contributed to the accumulation of flavonoids in the fruit ripening of the two main Actinidia arguta Sieb. Zucc. cultivars. Across many plants, naringin, kaempferol, kaempferol, myricetin, dihydromyricetin, and dihydro quercetin are catalyzed by positively regulating the expression of CHS, CHI, F3H, F3’H, and FLS (Liu et al., 2002; Deavours and Dixon, 2005; Liu et al., 2016; Wang et al., 2017; Matsui et al., 2018; Wang et al., 2019a; Jin et al., 2022; Li et al., 2022b). In addition, CHS, CHI, F3H, CFoL, LOC, LSAT, FNSI, DFR, F3’H, FLS, and HIDH may play structural or regulatory roles in the biosynthetic pathways of the flavonoid compounds chrysin (Flavone1), rutin (Flavone2) and daidzein (Flavone3). DFR, F3’H, and FLS compete with the substrate dihydrokaempferol (DHK). FLS expression and the encoded enzyme activity were high, which ultimately promotes Rutin accumulation. LOC expression and the levels of Chrysin and Daidzein were relatively low. In conclusion, the research results will continue to expand our further development of Actinidia arguta Sieb. Zucc. resources in northern China, and will further promote our exploration of the molecular basis of using flavonoids in Actinidia arguta Sieb. Zucc. to prevent and control hyperuricemia and its biosynthesis.
The data presented in the study are deposited in the NCBI repository, accession number PRJNA649743.
The animal study was reviewed and approved by Shenyang Agricultural University Institutional Animal Care and Use Committee (IACUC).
LC and YW designed and drafted the manuscript. YBW performed experiments and analyzed the data. YBW wrote the first draft of the manuscript with the help of KD and YW. XY has been involved in partial data analysis and figure compile and edit. MZ, CH and MI helped to review and edit the manuscript. LC has been involved in critically revising the manuscript for important intellectual content. All authors contributed to the article and approved the submitted version.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647163 | Yujiao Zhang,Hongyun Xing,Haoran Wang,Lan Yu,Zhi Yang,Xiangnan Meng,Pengpeng Hu,Haiyan Fan,Yang Yu,Na Cui | SlMYC2 interacted with the SlTOR promoter and mediated JA signaling to regulate growth and fruit quality in tomato | 27-10-2022 | Solanum lycopersicum,JA signaling,SlMYC2,TOR signaling,growth and fruit quality | Tomato (Solanum lycopersicum) is a major vegetable crop cultivated worldwide. The regulation of tomato growth and fruit quality has long been a popular research topic. MYC2 is a key regulator of the interaction between jasmonic acid (JA) signaling and other signaling pathways, and MYC2 can integrate the interaction between JA signaling and other hormone signals to regulate plant growth and development. TOR signaling is also an essential regulator of plant growth and development. However, it is unclear whether MYC2 can integrate JA signaling and TOR signaling during growth and development in tomato. Here, MeJA treatment and SlMYC2 overexpression inhibited the growth and development of tomato seedlings and photosynthesis, but increased the sugar–acid ratio and the contents of lycopene, carotenoid, soluble sugar, total phenol and flavonoids, indicating that JA signaling inhibited the growth of tomato seedlings and altered fruit quality. When TOR signaling was inhibited by RAP, the JA content increased, and the growth and photosynthesis of tomato seedlings decreased, indicating that TOR signaling positively regulated the growth and development of tomato seedlings. Further yeast one-hybrid assays showed that SlMYC2 could bind directly to the SlTOR promoter. Based on GUS staining analysis, SlMYC2 regulated the transcription of SlTOR, indicating that SlMYC2 mediated the interaction between JA and TOR signaling by acting on the promoter of SlTOR. This study provides a new strategy and some theoretical basis for tomato breeding. | SlMYC2 interacted with the SlTOR promoter and mediated JA signaling to regulate growth and fruit quality in tomato
Tomato (Solanum lycopersicum) is a major vegetable crop cultivated worldwide. The regulation of tomato growth and fruit quality has long been a popular research topic. MYC2 is a key regulator of the interaction between jasmonic acid (JA) signaling and other signaling pathways, and MYC2 can integrate the interaction between JA signaling and other hormone signals to regulate plant growth and development. TOR signaling is also an essential regulator of plant growth and development. However, it is unclear whether MYC2 can integrate JA signaling and TOR signaling during growth and development in tomato. Here, MeJA treatment and SlMYC2 overexpression inhibited the growth and development of tomato seedlings and photosynthesis, but increased the sugar–acid ratio and the contents of lycopene, carotenoid, soluble sugar, total phenol and flavonoids, indicating that JA signaling inhibited the growth of tomato seedlings and altered fruit quality. When TOR signaling was inhibited by RAP, the JA content increased, and the growth and photosynthesis of tomato seedlings decreased, indicating that TOR signaling positively regulated the growth and development of tomato seedlings. Further yeast one-hybrid assays showed that SlMYC2 could bind directly to the SlTOR promoter. Based on GUS staining analysis, SlMYC2 regulated the transcription of SlTOR, indicating that SlMYC2 mediated the interaction between JA and TOR signaling by acting on the promoter of SlTOR. This study provides a new strategy and some theoretical basis for tomato breeding.
Tomato (Solanum lycopersicum) is a major vegetable crops, and the regulation of its growth and development, as well as yield and quality, have been popular research topics. JA is not only an important plant growth regulator but also part of an important hormone signaling pathway, playing a key role in plant growth and development (Huang et al., 2017). The exogenous application of JA inhibits various aspects of seedling growth, including primary root growth, leaf expansion, and hypocotyl elongation (Wasternack and Hause, 2013; Song et al., 2014; Kim et al., 2015). Moreover, JA treatment inhibits the expansion of true leaves and cotyledons (Zhang and Turner, 2008; Chehab et al., 2012; Aleman et al., 2016). Important components of the JA signaling pathway include the protein COI1, complex SCFCOI1, protein JAZ, and transcription factor MYC2 (Chini et al., 2016; Li et al., 2021a; Song et al., 2022). MYC2 is the activating component of JA signaling and belongs to the transcription factor MYC basic helix loop helix (bHLH) IIIe subfamily. The MYC IIIe family in Arabidopsis consists of AtMYC2, AtMYC3, AtMYC4 and AtMYC5. However, in tomato, the MYC IIIe family consists of SlMYC1 and SlMYC2 (Heim et al., 2003; Chini et al., 2016; Goossens et al., 2016), and MYC2 is the main regulator of JA signaling activation (Yan et al., 2013; Xu et al., 2018; Liu et al., 2019; Min et al., 2020). MYC2 is widely present in animals and plants and has a variety of regulatory functions (Dombrecht et al., 2007; Altmann et al., 2020). In Arabidopsis thaliana, MYC2 directly binds to the promoters of PLT1 and PLT2, inhibiting their expression and thereby inhibiting the growth of the main roots (Chen et al., 2011). In rice, OsMYC2 regulates the expression of OsMADS1, a gene that regulates the development of floral organoids and activates the development of rice spikelets (Cai et al., 2014). In apple, MdMYC2 promotes the transcription of the ethylene synthesis genes MdACS1 and MdACO1, which in turn promotes ethylene synthesis in fruit and promotes fruit ripening (Li et al., 2017). In addition, MYC2 is also involved in complex metabolism. MYC2 plays an important role in regulating of artemisinin biosynthesis (Shen et al., 2016), and MYC2 regulates SGA biosynthesis in tomato and potato (Cárdenas et al., 2016; Thagun et al., 2016). The MYC2 protein contains a JID domain and an acidic AD domain at its N-terminus, and it contains a bHLH-zip domain and an ACT domain at its C-terminus (Thines et al., 2007). Elevated levels of JA-Ile promote the degradation of JAZ by the SCFCOI1 protease complex, followed by the transcriptional expression of MYC2, which initiates the expression of JA-responsive genes (Zhang et al., 2015; Breeze, 2019). This finding indicates that MYC2 plays an important role in JA-mediated plant metabolism and is a high-level transcriptional regulatory element in the JA signaling pathway. MYC2 positively regulates the inhibition of hypocotyl elongation by red or far-red light and negatively regulates the inhibition of hypocotyl elongation by blue light (Biswas et al., 2003; Riemann et al., 2008; Yan et al., 2012; Yang et al., 2012; Riemann et al., 2013), suggesting that MYC2 has different functions under different conditions. Moreover, MYC2 can integrate the interaction between JA signaling and other hormone signaling to regulate plant growth and development (Li et al., 2021b). One such example is for the ABA receptor PYL6, which interacts with the JA master regulator MYC2 to regulate the transcriptional activity of MYC2 (Aleman et al., 2016). JA interacts with various signaling pathways to mediate plant growth inhibition. JAs promote the degradation of JAZ to activate MYC2 and release DELLA protein (Hou et al., 2010; Yang et al., 2012). EIN3 is activated by the JA-mediated breakdown of JAZ proteins and ethylene (ET)-mediated stabilization. It binds to and represses MYC2. EIN3 also transcriptionally activates ORA59, and ORA59 represses MYC2 transcription. MYC2 can enhance its own transcription in the short term but repress it in the long term. EDS1 can repress MYC2 during AvrRps4-induced ETI. In addition, SA can promote JAZs degradation during ETI via NPR3 and NPR4. Generally, SA is an inhibitor of MYC2 transcription. Abscisic acid (ABA) directly activates the transcription of MYC2 and enhances the binding of the ABA receptor PYL6 to MYC2, which regulates MYC2 transcriptional activity, and MYC2 has different effects on the JAZ6 and JAZ8 promoters (Aerts et al., 2021). The above findings suggest that MYC2 mediates the interaction of JA signaling with other signaling pathways (Aerts et al., 2021). In both tomato and Arabidopsis, the active hormone JA-Ile promotes COI1-dependent degradation of the JAZ repressors and thereby activates the master TF MYC2. However, studies have found that the target genes downstream of MYC2 are different in tomato and Arabidopsis. In tomato, MYC2 positively and directly regulates the transcription of its downstream MTFs, which in turn regulates the expression of late wounding-responsive genes or pathogen-responsive genes. In contrast, in Arabidopsis, MYC2 positively regulates wounding-responsive genes while negatively regulating pathogen-responsive genes (Du et al., 2017). Target of rapamycin (TOR) is also an important signaling pathway in regulating plant growth and development and plays a central role in integrating metabolic energy and hormone signaling (Dobrenel et al., 2016). TOR is a highly conserved serine/threonine protein kinase in eukaryotes (Boutouja et al., 2019). Recent studies have found that the plant TOR complex can coordinate energy, growth, hormones and other signals (Zhuo et al., 2020; Fu et al., 2021), enabling it to regulate plant growth and development, as well as nutrition and energy processes, through the integration of its downstream effector proteins E2Fa and SPS (Xiong et al., 2017; De Vleesschauwer et al., 2018; Brunkard et al., 2020; Fu et al., 2020; O’Leary et al., 2020; Wang et al., 2020). Studies have shown that TOR signaling can regulate the growth and development of cotton fiber through the interaction of JAZ with JA signaling (Song et al., 2017). However, MYC2 is the core factor of signaling interactions, and it is unclear whether MYC2 can integrate JA signaling and TOR signaling pathways in response to the growth and development of tomato. Our study found that the presence of a cis-acting element of SlMYC2 on the SlTOR promoter, so it was speculated that SlMYC2 bound to the SlTOR promoter to regulate downstream response genes and thereby regulate the growth and development of tomato. In this research, we analyzed the roles of JA signaling, SlMYC2 and TOR signaling in tomato growth and development. We demonstrated that JA signaling inhibited the growth and development of tomato plants and altered fruit quality, and TOR signaling positively regulated the growth and development of tomato seedlings. In addition, yeast one-hybrid and GUS assays indicated that SlMYC2 could directly bind to the SlTOR promoter and positively regulate its transcription. These results suggested that MYC2 could integrate JA signaling and TOR signaling in respond to growth and development of tomato, providing a new strategy and some theoretical basis for tomato breeding.
Tomato seeds were heated in water at 55°C for 5 min, then soaked in 70% alcohol for 1 min in an ultra-clean bench, then soaked in 5% sodium hypochlorite for 5 min, washed with ddH2O, and then transferred into MS medium with sterile ophthalmic forceps. The substrate containing the seeds was placed in a light incubator and incubated for 3 d at 25/18°C protected from light; and then incubated under a light intensity of 20000 Lux, photoperiod of 16/8 h, and 25/18°C while germination. The germination rate, root length and hypocotyl length were observed and recorded. Using 0.1% DMSO as a control, the seedlings were treated with RAP (10 μM), MHY1485 (5 μM), DIECA (200 μM) and MeJA (100 μM), respectively.
The SlMYC2-OE and SlMYC2-RNAi lines were kindly provided by Dr. Chuanyou Li’s research group (Institute of Genetics and Developmental Biology Chinese Academy of Sciences). Transgenic lines of SlMYC2-OE and SlMYC2-RNAi were constructed using Gateway (Invitrogen) technology. A fragment of the SlMYC2 open reading frame (1–400 bp) was selected. The sequence was then cloned and inserted into pCAMBIA-1301 under the control of the CaMV35S promoter to generate the construct pCAMBIA-1301-SlMYC2-RNAi. For SlMYC2-OE tomato plants, the full-length coding sequence of SlMYC2 was amplified by PCR and cloned into the pGWB5 vector to generate the Pro35S:SlMYC2-GFP construct. The constructs were introduced into tomato cv M82 by Agrobacterium tumefaciens-mediated transformation. Transformants were selected based on their resistance to hygromycin. The A. tumefaciens-transformed T0 generation seeds were collected, and the T1 generation was screened by hygromycin and conformed to the ratio of 3:1, indicating that there were 1/3 homozygotes, and qRT−PCR was identified as homozygous and single copy. Then, the T1 generation was planted. After identification, 10 homozygous lines were selected, and the T2 generation of seeds from a single plant was collected. After hygromycin screening, no separation occurred, showing they were all homozygotes. The T3 seeds from the T2 generation were also collected, and the T4 seeds identified as homozygotes were collected through the same identification method as above. Homozygous seeds from T3 or T4 generation were all homozygotes after identification and screening. The homozygotes of T3 or T4 transgenic seedlings were used for phenotypic and molecular characterization (Du et al., 2017). Construction and transformation of tomato SlTOR gene silencing by virus-induced gene silencing (VIGS) method. A pTRV-based VIGS was performed to knock out the SlTOR in tomato cotyledons. A 402 bp fragment within the 3’-region of the SlTOR cDNA was cloned into the pTRV2 vector (TRV : SlTOR). The gene-specific primers were listed in Table S1 . 10 mL of each A. tumefaciens strain used was incubated overnight at 28°C in YEP medium supplemented with 100 mg·L-1 rifampicin and 50 mg·L-1 kanamycin. Then, 200 µL of each overnight culture was inoculated into a 20 mL portion of YEP medium containing the above antibiotics and incubated at 28°C until the culture reached the selected optical density of OD600 = 0.8-1.0. Induced A. tumefaciens strain EHA105 carrying different pTRV2-derived vectors (pTRV2 and pTRV2-SlTOR) was mixed with the pTRV1 A. tumefaciens strain EHA105 at a ratio of 1:1. The samples were spiked with 10 mM MES, 10 mM MgCl2, and 200 µM acetosyringone (AS) and then the germinated tomato seeds (root length = 0.2-0.6 cm) were immersed in the bacteria and vacuumed for 4 min. The seeds were sown and cultured in a growth chamber at 22°C with a 16 h light and 8 h dark photoperiod (Yu et al., 2019).
Fruit hardness was measured by a TMS-Pilot Precision texture analyzer (Food Technology Corporation, Virginia, USA), and the fruit color difference was measured by a color difference meter from CHROMA METER (KONICA MINOLTA SENSING INC. Japan). Lycopene was extracted with reference to Salvia-Trujillo and McClements (Salvia-Trujillo and McClements, 2016), total phenolic compounds were extracted with reference to Toor and Savage (Toor and Savage, 2005), and total flavonoids according to Jia et al. (Jia et al., 1999), while carotenoids were extracted by the acetone method and then all were determined spectrophotometrically. The contents of pectin, vitamin C, soluble protein, soluble sugar and acidity were also determined by spectrophotometrically.
High performance liquid chromatography (HPLC) was used to analyze the total JA content in the leaves of tomato plants.
The sequences of the SlTOR promoter were obtained from the NCBI website (http://www.ncbi.nlm.nih.gov/). The promoter sequences of SlTOR were analyzed by the promoter prediction website PlantCARE.
Total RNA from plants was extracted using TRIzol reagent (Gold Hi Plasmid Mini kit, CWO581M). BioDrop was used to measure the RNA concentration, and cDNA was synthesized by the FastKing cDNA First-Strand Synthesis Kit (catalog number: KR116) from Tiangen Biochemical Technology Co., Ltd. qRT−PCR was performed using a SYBR Green PCR Master Mix kit (TIANGEN, Beijing, China) on the CFX96 Touch Real-Time PCR Detection System (Bio-Rad). Primer sequences were listed in Table S1 . The reaction system (10 μL) contained 4.5 μL of 2× SuperReal PreMix Plus, 1 μL of cDNA, 0.5 μL of forward primer, 0.5 μL of reverse primer, and finally ddH2O was added to 10 μL. The reaction program included 95°C for 15 min and 40 cycles of 95°C for 10 s and 60°C for 32 s. Dissolution curve program included 65°C for 5 s and 95°C for 5 min.
The CDS of SlMYC2 and the 2000 bp promoter region of SlTOR were cloned using 2×Super Pfx MasterMix from Comway Century Biotechnology Co., Ltd. The SlMYC2 was then cloned into the pRI-101-GFP vector driven by the 35S promoter to obtain the SlMYC2 overexpression vector, and then the empty vector pRI-101-GFP and the recombinant vector SlMYC2-GFP were transformed into A. tumefaciens EHA105, respectively. The SlTOR promoter was cloned into the pRI-101-GUS vector to obtain a reporter gene driving GUS expression, and this vector and the recombinant vector were transformed into A. tumefaciens. The bacterial broth was inoculated into YEP liquid medium and incubated at 28°C with shaking at 200 rpm until OD600 = 0.8-1.0. Took 20 mL of different bacterial solutions into a centrifuge tube and centrifuged at 5000 rpm for 10 min, and discarded the supernatant. Resuspended and washed the bacteria with 2 mL of immersion solution (containing 10 mM MES and 10 mM MgCl2·6H2O) and centrifuged at 5000 rpm for 10 min, and discarded the supernatant. Resuspended cells in the soaking solution (containing 10 mM MES, 10 mM MgCl2·6H2O, and 200 mM AS) to make different combinations of cells with OD600 ≈ 1 and incubated at 28°C in the dark for 3 h. The reporter and effector were then mixed together in a 1:1 volume ratio to transform 5-week-old Nicotiana benthamiana. The empty pRI-101-GUS and pRI-101-GFP vectors served as controls. Primer sequences were listed in Table S1 . The different combinations of bacterial solutions were injected into tobacco leaves using a needleless syringe and incubated overnight in the dark, then removed and incubated normally for 1 d. The injected tobacco leaves (taken for 4 d consecutive) were immersed in 20 mL of GUS staining solution, vacuum-treated in the dark for 10 min, and stained in the dark at 37°C for 24 h. The tobacco leaves were decolorized to transparency and scanned for analysis by a scanner. Meanwhile, 0.5 g of injected tobacco leaves was placed in a 5 mL precooled centrifuge tube and set aside in liquid nitrogen. The leaves were ground to a powder using a precooled grinder, added to 2 mL of protein extract, mixed thoroughly and refrigerated at 4°C. The mixture was incubated for 1 h to fully react and then centrifuged at 2000 × g for 20 min at 4°C, removed the supernatant to a new centrifuge tube and centrifuged at 4°C and 2000 × g for another 10 min. Then 100 μL of the total protein supernatant was collected into a 1.5 mL centrifuge tube, and 900 μL of 37°C pretreated hot GUS extraction buffer was added. The tube was placed in a water bath at 37°C for 30 min before, added 200 μL to 800 μL of 0.2 M Na2CO3. The fluorescence intensity was measured using a fluorescence spectrophotometer at EM WL of 365 nm, EX WL of 455 nm and a slit of 3 nm.
The pGADT7 vector was digested with EcoR I and BamH I, and pAbAi was digested by Kpn I and Sal I. The carrier fragment was then recovered. Finally, the enzymatically cleaved vector fragment and the target fragment were ligated by one-step ligation (see Table S1 for the primer sequence list). The motifs of SlTOR promoter were analyzed using PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html). Based on the analysis results, the SlTOR promoter with a fragment size of 1856 bp was divided into three overlapping fragments, and each fragment overlapped by about 50 bp to avoid destroying the binding sites. The SlTOR promoter with a fragment size of 400-900 bp was cloned using 2×Super Pfx MasterMix (CWBIO, Jiangsu, China), and then promoters of SlTOR were inserted into the pAbAi vector to construct the baits (pAbAi-PSlTOR1/2/3 ). The pAbAi-PSlTOR was introduced into Y1H Gold, encapsulated on SD/-Ura medium plates, and incubated for 2-4 d at 30°C in inverted position. The plaque was identified by PCR. The positive plaques were recoated on a medium plate of SD/-Ura containing an AbA gradient (50-500 ng·mL-1). The pAbAi vector with promoter was screened for AbA resistance, i.e., the lowest concentration of AbA without phage spots was screened for subsequent yeast one-hybrid assays. AD-SlMYC2 was transformed into a yeast strain containing the pAbAi-PSlTOR linear vector and then encapsulated on SD/-Leu medium containing with AbA at the resistance concentration, and incubated upside down at 30°C to observe plaque growth.
SlMYC2-OE tomato seedlings grown to 15 d of age were used for ChIP−qPCR analysis. ChIP−qPCR assays were performed as described by Du (2017) (Du et al., 2017). The ChIP signal was analyzed using qPCR. Each ChIP value was normalized to its respective input DNA value, and the enrichment of DNA was displayed as a percentage of the input. Primer sequences were listed in Table S1 .
All experiments in this study were performed at least three repetitions. Experimental data were processed using Microsoft Excel 2010 and GraphPad Prism 6 for graphing. The significance of differences was determined by ANOVA or Student’s t test using IBM SPSS 20 software (P< 0.05). The data in the graphs were the mean ± SD (standard deviation).
JA is not only an important growth regulator but also an important signaling molecule for tomato growth and development. To investigate the regulatory effect of JA on growth and development, tomato seedlings were treated with MeJA (100 μM) and DIECA (200 μM. DIECA, JA biosynthesis inhibitor sodium diethyldithiocarbamate) (Ding et al., 2021). The results showed that MeJA inhibited plant height, stem thickness and dry weight ( Figures 1A–E ). Photosynthesis is closely related to the dry matter accumulation of plants. The products of photosynthesis are the basis for plant growth and development, the formation of yield and quality. Therefore, the photosynthetic indexes of 5-week-old tomato seedlings were examined using a photosynthesis meter Li-6400XT. MeJA treatment significantly reduced the net photosynthetic rate (Pn) and chlorophyll a and b contents of tomato seedlings ( Figures 1F–H ). The root–shoot ratio increased slightly after MeJA treatment, but the difference was not significant compared with the control ( Figure 1I ). In addition, for all parameters, there were no significant differences between the DIECA treated and control groups ( Figure 1 ).
SlMYC2 is a key transcription factor in the JA signaling pathway. To confirm the effect of SlMYC2 on the growth and development of tomato seedlings, we first examined the expression of SlMYC2 after MeJA and DIECA treatment. SlMYC2 expression was significantly enhanced after MeJA treatment. The expression of SlMYC2 decreased slightly in DIECA treatment, but it was not significant compared with the control, indicating that JA signaling could activate the expression of SlMYC2 ( Figure S1 ). Then, SlMYC2 overexpression and silenced lines were used as experimental materials ( Figure S2 ). The results indicated that the plant height, stem thickness and dry weight of the SlMYC2 overexpression lines were significantly lower than those of WT, but there were no significant differences between SlMYC2-silenced lines and WT ( Figures 2A–E ). In the SlMYC2 overexpression lines, the Pn and chlorophyll a and b contents were lower than those in the WT. Among them, Pn was remarkably lower in the overexpression lines than in WT, but the chlorophyll a and b contents were not significant. The Pn and chlorophyll a and b contents in the SlMYC2-silenced lines showed an increasing trends; however, the chlorophyll b content was observably higher than that in the WT, while the Pn and chlorophyll a content were not significant ( Figures 2F–H ). Nevertheless, the root–shoot ratio significantly increased in the SlMYC2 overexpression lines, but the effect was not significant in the SlMYC2-silenced lines, suggesting that SlMYC2 overexpression could enhance the number of healthy tomato seedlings ( Figure 2I ). The above results indicated that JA and SlMYC2, key transcription factor in the JA signaling pathway, had a regulatory role in the growth and development of tomato seedlings and the net photosynthetic rate.
Tomato fruit quality is an important indicator for its commercialization. Therefore, the relevant quality indicators of mature tomato fruits were determined after MeJA and DIECA treatment ( Figure S3 ). MeJA and DIECA treatment had no significant effect on tomato fruit firmness, indicating that their effects on fruit firmness were not obvious ( Figure S3A ). Although MeJA treatment significantly reduced the tomato fruit color index and total pectin ( Figures S3B, C ), it significantly increased the contents of lycopene, carotenoid, fructose, glucose, soluble sugar, starch, soluble protein, total phenol and flavonoids ( Figures S3D–H, K, L, N, O ), while the sugar–acid ratio also increased significantly ( Figure S3J ). MeJA treatment had no significant effects on total fruit acid or vitamin C content ( Figures S3I, M ). DIECA treatment only significantly improved the content of total pectin, but significantly decreased the content of flavonoids, however, while not significantly affecting other indicators ( Figure S3 ). This result suggested that increasing JA could improve the quality of tomato fruits. SlMYC2 is a key transcription factor in the JA signaling pathway; therefore, SlMYC2 overexpression and SlMYC2-silenced lines were also used to determine tomato fruit quality ( Figure 3 ). Tomato fruit firmness was significantly enhanced in the SlMYC2 overexpression lines but significantly reduced in the silenced line ( Figure 3A ). Tomato fruit color index was significantly lower in the SlMYC2 overexpression lines, but there was no significant difference between SlMYC2-silenced lines and WT ( Figure 3B ). The color change of fruit pulp is mainly determined by the accumulation of carotenoids and lycopene. Therefore, the lycopene and carotenoids contents in tomato fruits of each group were further determined. The lycopene and carotenoid contents in mature fruits were significantly higher in the SlMYC2 overexpression lines compared with those in the WT, but there was no significant difference in the SlMYC2-silenced lines ( Figures 3D, E ). The above results showed that SlMYC2-OE could increase the accumulation of lycopene and carotenoids and regulate fruit pulp color in mature tomato fruits. The effects of SlMYC2 on tomato fruit were analyzed by measuring the nutrient composition of ripe fruit from the SlMYC2 overexpression lines and SlMYC2-silenced lines. Total pectin content in ripe fruits from the SlMYC2 overexpression lines was decreased significantly, whereas it increased significantly in the ripe fruits from SlMYC2-silenced lines ( Figure 3C ). SlMYC2 overexpression significantly increased the contents of fructose, glucose, and soluble sugar but decreased the fruit acid content, whereas SlMYC2 silencing had no significant effect ( Figures 3F–I ). However, SlMYC2 overexpression increased the sugar–acid ratio in mature tomato fruits but decreased sugar–acid ratio in mature fruits of SlMYC2-silenced lines ( Figure 3J ). The contents of starch and total phenol in mature fruits of the SlMYC2 overexpression lines were significantly higher than those of the control ( Figures 3K, N ), whereas the content of total vitamin C was increased with SlMYC2 silencing ( Figure 3M ). The contents of soluble protein and flavonoids were significantly higher in mature fruits of SlMYC2 overexpression lines, and the content of flavonoids was significantly lower in the SlMYC2-silenced line than in the control group, but the difference in soluble protein content was not significantly different ( Figures 3L, O ). These results revealed that JA signaling and the key transcription factor SlMYC2 were involved in the regulation of tomato fruit quality.
TOR is conserved in eukaryotes and is one of the most important and highly conserved regulators of growth and development (Xiong et al., 2013). Three concentrations of RAP (SlTOR inhibitor) (1 μM, 5 μM and 10 μM) were employed to measure the germination rate and seedling height under different treatments. The results showed that the germination rate of tomato seeds and plant height were reduced by RAP treatment ( Figures S4A, B ). The 10 μM RAP concentration was superior to other concentrations, so 10 μM RAP was selected for further testing. Three MHY1485 (SlTOR activator) concentrations (1 μM, 5 μM and 10 μM) were tested. The results indicated that all concentrations could promote seed germination and seedling growth ( Figures S4C, D, E ) and accelerate relative hypocotyl length and root length ( Figures S4F, G ). However, overall, the 5 μM MHY1485 concentration was superior to other concentrations. Therefore, subsequent experiments were carried out using 5 μM MHY1485. The treatments with RAP (10 μM) and MHY1485 (5 μM) were used for further experiments. The results demonstrated that RAP inhibited the growth of tomato seedlings, but promoted by MHY1485 ( Figure 4A ). Moreover, the seed germination, relative hypocotyl elongation and root elongation of tomato were significantly inhibited by MeJA treatment but promoted by DIECA ( Figures 4B–G ). The photosynthetic changes of tomato seedlings were measured by a photosynthesis-testing instrument. When tomato seedlings were treated with RAP (10 μM), chlorophyll a and total chlorophyll ( Figures 5A, C ), the net photosynthetic rate (Pn) ( Figure 5D ) and the transpiration rate (Ts) ( Figure 5E ) of leaves were significantly reduced, while the changes in chlorophyll b ( Figure 5B ) and stomatal conductance (Gs) ( Figure 5F ) were not significant; however, the intercellular carbon dioxide concentration (Ci) was prominently increased ( Figure 5G ). The results suggested that the photosynthesis of tomato seedlings was inhibited after that RAP inhibited the TOR signaling pathway, but this change was caused by non-stomatal factors. Under the synergistic treatment of MeJA and RAP, the TOR inhibitor RAP and jasmonic acid MeJA synergistically inhibited the growth of tomato seedlings, and the hypocotyl length and root length were significantly reduced ( Table 1 , Figure S5 ). Further study showed that TOR affected the synthesis of endogenous JA. The content of JA in tomato seedlings was determined after treatment with the TOR inhibitor RAP and activator MHY1485. The results showed that RAP treatment increased the content of endogenous JA, but MHY1485 treatment significantly reduced the amount of JA ( Figure 6A ), indicating that TOR negatively regulated the synthesis of endogenous JA. Then, RAP and MeJA treatment ( Figure 6B ) and TRV : SlTOR lines ( Figure S6 ) were used to identify the expression of SlMYC2 gene and downstream response genes in the JA signaling pathway. The results showed that SlMYC2, TomLoxD (JA synthesis gene) and SlJA2L (SlMYC2 downstream target gene) were significantly upregulated after TOR signaling was suppressed by RAP and SlTOR silencing ( Figures 6B, C ). They were also upregulated by increasing MeJA, but SlTOR expression was downregulated by RAP and MeJA treatment, suggesting an interaction between TOR and JA signaling, which jointly regulated the growth of tomato seedlings.
Further bioinformatic analysis revealed that there were several predicted MYC2 binding elements existed in the SlTOR promoter ( Figure 7A and Table S2 ), suggested that SlMYC2 might bind to the promoter of SlTOR and activate its expression in vivo. The binding ability of SlMYC2 to SlTOR promoter was firstly verified by yeast one-hybrid (Y1H) assay. In Y1H assay, according to the positions of the predicted elements, we divided the SlTOR promoter into three fragments: PSlTOR1 , PSlTOR2 and PSlTOR3 ( Figure 7A ). The results showed that yeast cells could growth under 550 ng·μL-1 AbA, indicating a potential interaction between SlMYC2 and PSlTOR3 . However, pAbAi-PSlTOR1 and pAbAi-PSlTOR2 were self-activated under the same AbA concentration ( Figure 7B and Figure S7 ). To further confirm SlMYC2 bound to the SlTOR promoter in vivo, we performed chromatin immunoprecipitation (ChIP) experiments using SlMYC2-OE seedlings. We divided the SlTOR promoter into 7 segments for ChIP experiments ( Figure 7A ), and the details of the segments were shown in the Figures 7A and Table S2 . The ChIP results showed that each fragment of the SlTOR promoter was enriched in some degree, but highest enrichment occured at P5 segment which located in PSlTOR3 ( Figure 7C ), suggesting that potential binding site of SlMYC2 to SlTOR promoter might locate in -686 to -826 of the ATG upstream. Both Y1H and ChIP assays suggested that SlMYC2 bound to SlTOR promoter directly. To assay the activation effect of SlMYC2 on SlTOR transcription in vivo, transient transfections with different reporter and effector vectors were performed respectively. In the GUS histochemical staining assay, transformations with PSlTOR ::GUS and 35S::GFP were used as controls ( Figures 7D, E ). The staining results showed that GUS signal could not been observed when transformed with PSlTOR ::GUS and 35S::GFP; however, it was appeared when co-transformed with both PSlTOR ::GUS and SlMYC2::GFP ( Figure 7E ). GUS enzyme activity analysis also supported the staining result. The GUS activity was significantly higher than those of the controls ( Figure 7E ), suggesting that SlMYC2 regulated the transcription of SlTOR. Moreover, we also found that the expression of SlTOR was obviously upregulated in the SlMYC2 overexpression line ( Figure 7F ). Taken together, SlMYC2 might mediate the crosstalk between the JA and TOR signaling pathways through directly binding to the SlTOR promoter and activating its expression.
The growth and development of seedlings are of great importance for obtaining vigorous seedlings and improving yield, quality and resistance. The growth and development of plants are regulated by many plant hormones. Recent studies have shown that JA plays an important role in the regulation of plant growth and development. The phytohormone JA regulates a wide range of biological processes (Chini et al., 2016; Huang et al., 2017; Li et al., 2021a; Song et al., 2022). Jasmonates (JAs), which consist of jasmonic acid and its derivatives, such as methyl jasmonate (MeJA) and jasmonoyl-isoleucine, play multiple roles in growth and development as well as biotic and abiotic stress responses. Numerous studies have shown that JA is involved in root hair formation, stamen development, flowering, leaf senescence, anthocyanin biosynthesis and photosynthetic carbon fixation (He et al., 2002; Li et al., 2014; Qi et al., 2015; Ding et al., 2018; Wang et al., 2019; Han et al., 2020). JA is both a hormone and a signaling molecule. MYC2 is a master regulator in the JA signaling pathway that is widely present in plants and animals (Zhang et al., 2015; Breeze, 2019) and has multiple regulatory functions (Chini et al., 2007; Sheard et al., 2010; Li et al., 2021b). MYC2 is an essential helix−loop−helix transcription factor that plays a critical role in the JA signaling pathway and is involved in a variety of plant growth and stress resistance processes. However, the effects of activation and inhibition of JA signaling, and enhancement and silencing of SlMYC2 on the growth and photosynthesis of tomato seedlings, especially the regulation of tomato fruit quality, have not been systematically investigated. The formation of fruit quality is inseparable from the growth of tomato seedlings. Only when the plants maintain continuous and healthily growth, can they provide sufficient nutrients for normal fruit development and maturation. To investigate the role of JA and its key transcription factor MYC2 in the growth of tomato seedlings, MeJA and DIECA were used as treatments, and SlMYC2-OE and SlMYC2-RANi lines were used to measure the growth and developmental indexes of tomato seedlings. The results showed that MeJA treatment and the SlMYC2 overexpression inhibited the growth, development and photosynthesis of tomato seedlings but increased the root–shoot ratio. Increasing MeJA or the overexpression of SlMYC2 inhibited the growth of tomato seedlings to some extent, but resulted in vigorous seedlings, indicating that JA signaling played a role in the growth and development of tomato seedlings. Some studies showed that application of MeJA during preharvest and postharvest stages could enhance fruit antioxidant capacity and phenolic content, thereby extending the shelf life and improving fruit quality (Tzortzakis, 2007; Liu et al., 2018). The insignificant variation in lycopene content in fruits of lines with different jasmonic acid content indicates that the regulation of lycopene accumulation by jasmonic acid may be dependent on specific concentrations (Liu et al., 2012). Some experiments also show that the content of lycopene in the fruit of the JA synthetic mutant spr2 decreases, suggesting a positive correlation between endogenous JA content and lycopene content in fruits. Exogenous application of MeJA can restore the lycopene content in mutant fruits (Min et al., 2018). Using the SlMYC2-silenced tomato fruit obtained by VIGS technology as the material, the results also show that SlMYC2 plays a critical role in MeJA-induced fruit chilling resistance, including the components related to fruit quality (Min et al., 2018). The above evidence suggests that JA and MYC2 play important roles in the regulation of tomato fruit quality. However, systematic studies on the changes of JA and SlMYC2 on tomato fruit quality are lacking. In this research, we systematically studied the effects of JA and MYC2 on the quality of mature tomato fruits. The results showed that the contents of lycopene and carotenoid increased significantly after MeJA (100 μM) treatment and SlMYC2 overexpression. Moreover, the contents of soluble sugar, protein, total phenol and flavonoids were also increased. MeJA processing and SlMYC2 overexpression increased the sugar–acid ratio to improve the quality of fruits, and SlMYC2 overexpression enhanced fruit firmness. Increasing MeJA or the overexpression of SlMYC2 improved the quality of tomato fruits, and overexpression of SlMYC2 increased fruit firmness and prolonged shelf life, indicating that JA signaling played an important role in tomato fruit quality.
MYC2 is the key transcription factor in the JA signaling transduction pathway and is a critical component of JA signaling that interacts with other signaling pathways (Dombrecht et al., 2007; Altmann et al., 2020). MYC2 can integrate the interaction between JA signaling and other signals to regulate plant growth and development (Breeze, 2019). Target of rapamycin (TOR) is also an important signaling pathway that regulates plant growth and development, playing a central role in integrating metabolic energy and hormone signals (Dobrenel et al., 2016; Li et al., 2022). Through bioinformatic analysis, the binding elements of MYC2 were identified in the promoter of SlTOR. Therefore, it was hypothesized that SlMYC2 might be the node that mediated the crosstalk between the JA and TOR signaling pathways. Studies have shown that JA signaling and TOR signaling are the core signals that regulate plant growth and development, such as seed germination, rhizome elongation, tricarboxylic acid cycle, starch storage and fruit quality (Xiong et al., 2013; Okello et al., 2015; Song et al., 2017). In addition, there may be an interaction between JA signaling and TOR signaling (Göbel and Feussner, 2009; Mosblech et al., 2009; Salem and Giavalisco, 2019). To clarify the role and relationship between JA and TOR signaling in tomato growth and development, tomato seedlings were treated with MHY1485 (an activator of TOR) and RAP (an inhibitor of TOR). When TOR signaling was inhibited by RAP, it increased the level of JA and inhibited the growth and photosynthesis of tomato seedlings, indicating that TOR signaling positively regulated the growth and development of tomato seedlings. Inhibition of TOR signaling and activation of JA signaling synergistically inhibited the tomato germination and seedling growth; that is, JA signaling and TOR signaling might be involved in crosstalk to regulate the growth and development of tomato seedlings. This result was consistent with previous findings (Song et al., 2017). Moreover, endogenous JA content was significantly reduced after activation of TOR signaling by MHY1485. The endogenous JA content increased after RAP inhibition of TOR, suggesting that TOR signaling inhibited JA regulation during plant growth and development by suppressing JA synthesis, and that SlMYC2 might mediate the crosstalk between JA signaling and TOR signaling pathways. To show the effect of SlMYC2 on SlTOR transcription, the transcription of SlTOR in SlMYC2-silenced and overexpression lines was examined. The results showed that SlMYC2 promoted the accumulation of SlTOR transcription. Therefore, we tentatively determined that SlMYC2 could affect SlTOR signaling by regulating SlTOR transcription. To explore the mechanism by which SlMYC2 regulates SlTOR transcription, GUS staining, GUS enzyme activity and yeast one-hybrid assays were employed. In this study, yeast one-hybrid and ChIP assays obtained the same results that SlMYC2 could directly interact with SlTOR. Further analysis found that the highest enrichment occurred at P5 segment which located in PSlTOR3 , and the PSlTOR3 contained the CACGTG and CACATG motifs in P5 segment, suggesting that potential binding sites of SlMYC2 in SlTOR promoter might be located in -686 to -826 of the ATG upstream. It was determined that SlMYC2 could activate the transcriptional accumulation of SlTOR by binding to the SlTOR promoter. This interaction might be one of the reasons why SlTOR regulated JA signaling by regulating SlMYC2. The transcription factor MYC2 is a key factor in JA accumulation and signaling activation, and its regulation necessarily induces the response to JA signaling pathway (Heim et al., 2003; Chini et al., 2016; Goossens et al., 2016). SlTOR may regulate the response to the corresponding genes of JA by regulating the level of SlMYC2. Therefore, TRV-SlTOR lines were constructed to detect the expression of JA synthesis genes and SlMYC2 target genes (TomLoxD and SlJA2L). The results showed that the expression of SlMYC2 and its target genes was significantly upregulated. The activation of SlMYC2 and its target genes were significantly enhanced when SlTOR signaling pathway was inhibited, which uncovered that SlMYC2 mediated the interaction between JA and TOR signaling by acting on the promoter of SlTOR. Studies in Drosophila and mammals have shown that TOR can activate PP2A to dephosphorylate MYC, leading to a decrease in MYC stability and protein content. When mTOR is inhibited, MYC2 synthesis is blocked and MYC2 transcription is enhanced (Lee et al., 2017; Shen et al., 2018). However, more plant evidence is needed. In our unpublished data, the accumulation of SlMYC2 target genes and their protein content were significantly activated by the inhibition of TOR signaling. Therefore, subsequent studies will further reveal whether SlTOR regulates JA signaling through SlMYC2. In conclusion, SlMYC2 overexpression could significantly inhibit the growth of tomato seedlings but could improve the quality of tomato fruits. Biochemical experiments showed that inhibition of TOR signaling promoted JA synthesis in tomato and increased the amount of JA in the plant, releasing SlMYC2. SlMYC2 combined with the SlTOR promoter, thus activating the SlTOR gene, which might feedback on the inhibition of SlMYC2 ( Figure S8 ). However, the feedback regulation of SlMYC2 by TOR signaling and the effect on JA signaling still need to be further explored. JA is an anti-stress hormone. The level of JA is low in plants under normal conditions, but under stress, JA is elevated to initiate SlMYC2 and activates SlTOR in combination with the SlTOR promoter; thus, feedback inhibits JA synthesis and SlMYC2 expression, leading to a balance of growth and stress resistance. Therefore, the present study can lay the foundation for further revealing the mechanism of JA and TOR signaling interactions in regulating plant growth and development as well as stress resistance balance under stress.
The original contributions presented in the study are included in the article/ Supplementary Materials . Further inquiries can be directed to the corresponding authors.
YZ and HX contributed to the experimental design, tomato planting and sampling, tobacco planting and experimentation, data processing and result analysis and writing. HW, LY, ZY, and XM contributed to the analysis of the data in the experiment. PH, HF, YY, and NC revised the paper. All authors contributed to the article and approved the submitted version.
This research was funded by the National Key Research and Development Program Projects of China (2019YFD1000300) and the Key Project of Science and Technology Research of Liaoning Provincial Education Department (LJKZ0631).
We are particularly grateful to Dr. Chuanyou Li (Institute of Genetics and Developmental Biology Chinese Academy of Sciences) for kindly providing the SlMYC2 mutants. We would like to thank American Journal Experts (www.aje.com) for English language editing.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647175 | Marisa Conte,Daniela Eletto,Martina Pannetta,Anna M. Petrone,Maria C. Monti,Chiara Cassiano,Giorgio Giurato,Francesca Rizzo,Peter Tessarz,Antonello Petrella,Alessandra Tosco,Amalia Porta | Effects of Hst3p inhibition in Candida albicans: a genome-wide H3K56 acetylation analysis | 27-10-2022 | Candida,sirtuin,Hst3p deacetylase,H3K56 acetylation,ChIP-seq,antifungals | Candida spp. represent the third most frequent worldwide cause of infection in Intensive Care Units with a mortality rate of almost 40%. The classes of antifungals currently available include azoles, polyenes, echinocandins, pyrimidine derivatives, and allylamines. However, the therapeutical options for the treatment of candidiasis are drastically reduced by the increasing antifungal resistance. The growing need for a more targeted antifungal therapy is limited by the concern of finding molecules that specifically recognize the microbial cell without damaging the host. Epigenetic writers and erasers have emerged as promising targets in different contexts, including the treatment of fungal infections. In C. albicans, Hst3p, a sirtuin that deacetylates H3K56ac, represents an attractive antifungal target as it is essential for the fungus viability and virulence. Although the relevance of such epigenetic regulator is documented for the development of new antifungal therapies, the molecular mechanism behind Hst3p-mediated epigenetic regulation remains unrevealed. Here, we provide the first genome-wide profiling of H3K56ac in C. albicans resulting in H3K56ac enriched regions associated with Candida sp. pathogenicity. Upon Hst3p inhibition, 447 regions gain H3K56ac. Importantly, these genomic areas contain genes encoding for adhesin proteins, degradative enzymes, and white-opaque switching. Moreover, our RNA-seq analysis revealed 1330 upregulated and 1081 downregulated transcripts upon Hst3p inhibition, and among them, we identified 87 genes whose transcriptional increase well correlates with the enrichment of H3K56 acetylation on their promoters, including some well-known regulators of phenotypic switching and virulence. Based on our evidence, Hst3p is an appealing target for the development of new potential antifungal drugs. | Effects of Hst3p inhibition in Candida albicans: a genome-wide H3K56 acetylation analysis
Candida spp. represent the third most frequent worldwide cause of infection in Intensive Care Units with a mortality rate of almost 40%. The classes of antifungals currently available include azoles, polyenes, echinocandins, pyrimidine derivatives, and allylamines. However, the therapeutical options for the treatment of candidiasis are drastically reduced by the increasing antifungal resistance. The growing need for a more targeted antifungal therapy is limited by the concern of finding molecules that specifically recognize the microbial cell without damaging the host. Epigenetic writers and erasers have emerged as promising targets in different contexts, including the treatment of fungal infections. In C. albicans, Hst3p, a sirtuin that deacetylates H3K56ac, represents an attractive antifungal target as it is essential for the fungus viability and virulence. Although the relevance of such epigenetic regulator is documented for the development of new antifungal therapies, the molecular mechanism behind Hst3p-mediated epigenetic regulation remains unrevealed. Here, we provide the first genome-wide profiling of H3K56ac in C. albicans resulting in H3K56ac enriched regions associated with Candida sp. pathogenicity. Upon Hst3p inhibition, 447 regions gain H3K56ac. Importantly, these genomic areas contain genes encoding for adhesin proteins, degradative enzymes, and white-opaque switching. Moreover, our RNA-seq analysis revealed 1330 upregulated and 1081 downregulated transcripts upon Hst3p inhibition, and among them, we identified 87 genes whose transcriptional increase well correlates with the enrichment of H3K56 acetylation on their promoters, including some well-known regulators of phenotypic switching and virulence. Based on our evidence, Hst3p is an appealing target for the development of new potential antifungal drugs.
The polymorphic fungus Candida albicans is the major human opportunistic fungal pathogen (Lohse et al., 2018); its medical importance makes it a suitable experimental model for studying fungal pathologies and the underlying biology of dimorphic fungi (Kabir et al., 2012). The pathogenic cues for C. albicans are mostly environmental shifts such as a compromised immune system, a change of microbial flora following antibiotic treatment, or a medical device implant and organ transplantation (Lamoth et al., 2018). The ability of C. albicans to survive and proliferate in hostile environments (Mayer et al., 2013), combined with the dynamic morphological transition from yeast to filament form, is the most noticeable determinant for virulence rather than its static morphology, yeast or hypha, itself (Sudbery et al., 2004; Kadosh, 2016). Therefore, transcriptional regulation behind this dynamic morphology is directly linked to the control of C. albicans virulence. As in all eukaryotes, both transcriptional repression and activation are directly influenced by chromatin structures, whose dynamic change depends mainly on posttranslational histone modifications (Fallah et al., 2021). Indeed, many studies revealed that the epigenetic regulation by chromatin structure modifiers, along with well-known transcription factors in the signaling pathways, could be linked to the morphological phenotype transition (Klar et al., 2001; Simonetti et al., 2007; Zacchi et al., 2010; Rai et al., 2018). Specifically, the yeast to hyphae transition, the white-opaque switching, the ability to develop biofilm, and the adaptation to drug pressure are interesting and essential pathogenic mechanisms of this fungus, and posttranslational histone modifications play a prominent role in their regulation (Hnisz et al., 2009; Li et al., 2015; Kim et al., 2015; Garnaud et al., 2016; Qasim et al., 2021). Among various chromatin modifications, histone acetylation-deacetylation plays a leading role since it regulates pathogenic processes and promotes C. albicans virulence (Simonetti et al., 2007; Zacchi et al., 2010; Garnaud et al., 2016). In C. albicans, histone deacetylases (HDAC) have been divided into three groups: Class I and Class II HDACs, which are zinc-dependent, and Class III HDACs, namely the sirtuin family, which are NAD+ dependent (Su et al., 2020). Recent studies showed that sirtuins might be involved in sensing environmental changes triggering morphological transition (Zhao and Rusche, 2021). In C. albicans, there are five sirtuin-coding genes (SIR2, HST1, HST2, HST3, and orf19.2963) with similarities to Homo sapiens SIRT5. Sir2p and Hst1p are paralogs resulting from an ancient gene duplication, whereas Hst2p is evolutionarily more divergent (Rupert et al., 2016). However, they have distinct roles and different localization in C. albicans cells, as Sir2p localizes in the nucleolus, Hst1p in the nucleus, and Hst2p in the cytoplasm (Zhao and Rusche, 2021). In particular, Hst1p, a component of the Set3 complex (Set3C), is thought likely to repress yeast-to-filament transition by attenuating the cAMP/PKA signaling pathway while promoting the white-to-opaque switching (Su et al., 2020). On the other hand, a decrease in the expression of hypha-specific genes HWP1, ALS3, and ECE1 was observed in the sir2Δ/Δ mutant (Zhao and Rusche, 2021). In C. albicans, the NAD+-dependent histone deacetylase (sirtuin) Hst3p is responsible for removing the acetyl group of the Lys 56 of the H3 histone, which is particularly abundant in yeasts; it marks newly synthesized histones, facilitating their deposition onto chromatin and chromatin segments with high nucleosome turnover. This posttranslational modification is significant in yeasts because it regulates DNA damage response and contributes to fungal genome integrity (Celic et al., 2006; Wurtele et al., 2010). In C. albicans, H3K56 acetylation/deacetylation balance is regulated by the acetyltransferase (HAT) Rtt109p and Hst3p deacetylase, encoded by RTT109 and HST3 genes, respectively (Wurtele et al., 2010). Previous studies highlighted that HST3 deletion is lethal for C. albicans (Wurtele et al., 2010), suggesting that this is an essential gene in this fungus. In addition, HST3 conditional mutants displayed attenuated virulence in mice models; moreover, a reduced copy number of HST3, or inhibition of Hst3p by nicotinamide (NAM) treatment, promoted white-to-opaque switching with a consequent increased opaque phenotype (Wurtele et al., 2010; Stevenson and Liu, 2011; Stevenson and Liu, 2013; Guan and Liu, 2015). Like antibiotic-resistant bacteria, the occurrence of fungal strains resistant to the main antifungal drugs (including polyenes, azoles, echinocandins, and 5-Flucytosine) is becoming a serious threat to worldwide public health. Moreover, the available therapeutic molecules lack specificity and selectivity with consequent multiple side effects. Given the central role of histone acetylation/deacetylation balance in C. albicans growth and virulence, HATs and HDACs represent a promising target for developing new antifungal agents. Fungal enzymes that regulate H3K56ac levels diverge significantly from their human counterparts; indeed, fungal Hst family members share sequence motifs absent in human sirtuins (Wurtele et al., 2010). In addition, Hst3p has high substrate specificity, unlike the sirtuins Sirt1 and Sirt2 involved in H3K56 deacetylation in human cells (Celic et al., 2006). Based on this evidence, Hst3p represents a unique and interesting target for the development of new antifungals. Despite ongoing efforts over recent years, the roles played by Hst3p and H3K56 acetylation dynamics are only partially understood. Therefore, in this study, using the non-specific sirtuin inhibitor NAM, a product of NAD+-dependent deacetylation reaction, we evaluated the effect of Hst3p inhibition on the acetylation levels of its substrate H3K56 during yeast growth as well as on morphology and transcription profile of C. albicans. Moreover, by Chromatin immunoprecipitation followed by sequencing (ChIP-seq) analysis, we provide, for the first time, a genome-wide map of H3K56ac and identify some virulence-related genes whose transcription is directly or indirectly regulated by Hst3p.
NAM was purchased by Sigma-Aldrich (Milan, Italy). For all the experiments, 2 M NAM in ultra-pure distilled water was used as stock and added to cultures to obtain the required concentrations. Sirtinol, SirReal2, and Inauhzin, purchased from Selleckchem (Planegg, Germany), were dissolved in dimethylsulfoxide (DMSO) (Sigma-Aldrich, Milan, Italy).
C. albicans wild-type strain SC5314 (ATCC-MYA-2876) was routinely cultured on YPD (1% Yeast extract, 2% Peptone, 2% Dextrose) agar plates and propagated in liquid YPD medium overnight at 25°C at 200 rpm. Depending on experimental conditions, C. albicans was grown in synthetic YNB medium (0.17% Difco Yeast Nitrogen Base, without amino acids and ammonium sulfate), supplemented with 2% glucose and 0.5% ammonia sulfate; M199 medium containing Earle’s salts and glutamine (Sigma-Aldrich, Milan, Italy), buffered at pH 7.5 with 25 mM HEPES; 10% FBS (fetal bovine serum) (Euroclone, Pero, Italy); RPMI 1640 medium with L-Glutamine (Euroclone, Pero, Italy), Spider medium (2% mannitol, 2% nutrient broth, 0.4% K2HPO4, pH adjusted to 7.2 with NaOH). Each medium was supplemented with 10 or 25 mM NAM depending on the experimental needs. Solid media were prepared by adding 2% agar to the liquid broth before autoclaving. The optical density (OD) of each culture was measured at a wavelength of 600 nm (OD600). Three independent biological replicates were performed.
A budding yeast culture was diluted in 6-well plates to a density of 105 cell/ml in 2 mL YPD with 10 mM NAM (CaNAM), 50 µM Inauhzin, 10 µM Sirtinol or 50 µM SirReal2. Candida cells treated with only vehicle (DMSO) were used as control (CTRL). Plates were incubated at 30°C for 28 hours, and cell differentiation was followed by Time Lapse imaging (Leica DMI6000 T, Buccinasco, Italy). A total of 5 x 108 cells were harvested by centrifugation (4,700g, 10 min at 4°C), washed with distilled water, and then stored at -80°C for subsequent histone extraction. Three independent biological replicates were performed.
Over-weekend yeast cell culture was diluted to 107 cells/mL in 50 mL YPD medium and allowed to grow at 25°C overnight. Subsequently, the budding yeast culture was used to inoculate 700 mL YPD (with and without 10 mM NAM) at a cell density of 2x106/mL. Treated (CaNAM) and untreated yeast cells (CTRL) were cultured for 50 hours with orbital shaking (200 rpm) at 25°C, and growth was followed by measuring the OD600. At selected time points (0, 2, 4, 6, 8, 10, and 28 h), a pellet of 5 x 108 cells was collected by centrifugation (4700g, 10 min at 4°C), washed with distilled water, and then stored at -80°C for subsequent histone extraction.
C. albicans grown overnight in YNB at 25°C were collected by centrifugation (2,000g, 10 min at 25°C), washed with 0.15 M NaCl, resuspended in 0.15 M NaCl, and incubated at 25°C for 24 h to induce starvation. Subsequently, 106 yeasts/mL were inoculated in RPMI 1640 for 6 hours at 25, 30, or 37°C so that the cell morphology was regulated only by the growth temperature, in particular, obtaining yeast cells at 25°C, germ tubes at 30°C and true hyphae at 37°C. Cells were harvested by centrifugation (4700g, 10 min at 4°C), the pellets were washed with distilled water, and then stored at -80°C for subsequent histone extraction. Three independent biological replicates were performed.
C. albicans pellets were first resuspended in 10 mM EDTA, 10 mM Tris-HCl pH 7.4, 5 mM sodium butyrate, 5 mM NAM, 2.5% 2-Mercaptoethanol and 10% glycerol. The cellular suspension was ground to a fine powder with a mortar and pestle in liquid nitrogen. To enhance the lysis efficiency, the powder was resuspended in 10 mM EDTA, 10 mM Tris-HCl pH 7.4, 5 mM sodium butyrate, 5 mM NAM, 2.5% 2-Mercaptoethanol, 1% SDS, and 2% Triton-X 100 and acid-washed glass beads were added. Two or more cycles of freeze and vortex were performed. The suspensions were centrifuged at 12,000g for 15 min at 4°C, and the supernatants were collected. Acid soluble proteins were extracted from the total lysates by adding 0.4 N H2SO4 followed by incubation at 4°C for 3 h, under gentle inversion. Histones were acid-precipitated with 25% TCA (trichloroacetic acid, Sigma-Aldrich, Milan, Italy) overnight at 4°C. After two washing steps with ice-cold acetone, precipitated proteins were resuspended in Milli-Q H2O. Histones were resolved on SDS-PAGE (15% polyacrylamide gel), and the gel was stained with Coomassie G-250 Brilliant Blue (Sigma-Aldrich, Milan, Italy) and destained in water. For Western blotting, 0.5 μg of each histones sample were resolved on SDS-PAGE and transferred to a nitrocellulose membrane using the Trans-Blot Turbo Transfer System (Bio-Rad, Segrate, Italy). The following primary antibodies were used for detection: H3 (Abcam, ab1791, Cambridge, UK) and H3K56ac (Active Motif, Waterloo, Belgium). Band densities were visualized by LAS 4000 (GE Healthcare, Life Sciences) digital imaging system and quantified using ImageJ analysis software.
Histones H3 gel bands were cut from the Coomassie-stained gel and subjected to trypsin in situ digestion, as described by Shevchenko et al. (Shevchenko et al., 2006). Extracted peptides were dissolved in 10% formic acid before nano-ESI-LC-MS/MS analysis. Peptides were separated by a nano Acquity LC system (Waters Corp. Manchester, UK) equipped with a BEH C-18 1,7 µm, 75 µm x 250 mm (Waters Corp. Manchester, U.K) column connected to LTQ-Orbitrap hybrid mass spectrometer (Thermo Scientific). 5 μL of each sample were loaded onto the column and separated at a flow rate of 280 nL/min in a 15% - 40% buffer B linear gradient (Buffer A: 95% H2O, 5% ACN, 0.1% AA; Buffer B: 95% ACN, 5% H2O, 0.1% AA) in 55 minutes. Nano-ESI-LC-MS/MS analyses of H3 tryptic peptides were performed using Selected Reaction Monitoring (SRM) method on LTQ-Orbitrap mass spectrometer. The amount of acetylated peptide of interest (FQK(ac)STELLIR) was quantified by monitoring its bi-charged ion at m/z 638.82 and its fragmentation which produced two ions (831.54 and 1001.61). H3 peptide YKPGTVALR, an unmodified peptide, was used to normalize H3 quantities in each gel band, monitoring its bi-charged ion at m/z 502.86 and the fragmentation that produced two ions at m/z 616.38 and 713.50.
To evaluate morphologic changes induced by NAM treatment, C. albicans yeasts grown in YNB at 25°C were diluted to a cell density of 2 x 107 cells/mL, and 5 μL of this dilution were spotted onto solid media containing or not 25 mM NAM. In particular, YPD agar plates were incubated at 25°C, whereas 10% FBS, M199, and Spider medium were incubated at 37°C for hyphal growth induction. After 72 h of incubation, colony morphology was examined by microscopy at 10x magnification using an AMG Evos Imaging System (Thermo Fischer Scientific, Monza, Italy).
Budding yeast cell culture was diluted at a cell density of 105 cells/mL, distributed in Petri dishes of 90 mm diameter (10 mL each), and incubated at 25°C for 28 h with or without 10 mM NAM. Afterward, C. albicans cultures were cross-linked with 1% formaldehyde for 15 minutes at room temperature with gentle shaking. The reaction was quenched by adding 125 mM glycine and incubating for 5 min at room temperature under gentle shaking. Chromatin immunoprecipitation (ChIP) was performed as previously described (Mawer et al., 2021), except for cell lysis that was carried out by using a cryogenic freezer mill (SPEX SamplePrep 6970EFM Freezer/Mill, München, Germany). ChIP-seq libraries were generated from two independent biological replicates of H3K56ac and input following a previously published protocol (Ford et al., 2014) and sequenced on Illumina NextSeq 500 using 2 × 75 bp reads. Two independent biological replicates were performed for either control cells (CTRL) or C. albicans treated with 10 mM NAM (CaNAM).
As described above, C. albicans cultures were grown with or without 10 mM NAM in Petri dishes of 90 mm diameter and incubated at 25°C. Three independent biological replicates were performed for either CTRL or CaNAM cells. After 28 h of incubation, a total of 109 yeast cells were harvested (8,000g, 10 min at 4°C) and washed with diethylpyrocarbonate (DEPC)-treated water. RNA isolation was performed as described by Vennapusa et al., 2020 with some modifications. Briefly, yeast cells were resuspended in 600 mL of RNA extraction buffer (100 mM Tris- HCl (pH: 8), 25 mM EDTA, NaCl 2.5 M, 2.5% β-Mercaptoethanol, 2% SDS) and disrupted mechanically with BeadBug microtube homogenizer (Benchmark Scientific, Sigma-Aldrich, Milan, Italy) by using acid-washed glass beads. Total RNA was isolated using an equal volume of Acid-Phenol/Chloroform, pH 4.5 with IAA, 25:24:1 (Sigma-Aldrich, Milan, Italy). Following centrifugation, the RNA was precipitated from the aqueous phase with 100% ethanol, washed twice with 75% ethanol, and dissolved in DEPC-water. RNA quantification was carried out with the instrument Nanodrop 200 Thermo Fisher Scientific (Monza, Italy).
For RNA sequencing, RNA quality was assessed with TapeStation (Agilent, Cernusco sul Naviglio, Italy), and only RNA with RIN > 8 was used for library production. According to the manufacturer instructions, indexed libraries were prepared from 1 µg of purified RNA using TruSeq Stranded Total RNA Sample Prep Kit (Illumina Inc., Berlin, Germany). Libraries were pooled and sequenced (paired-end, 2 x 100 bp) on NextSeq 550 platform (Illumina Inc., Berlin, Germany). A Ribo-Zero Gold rRNA Removal Kit specific for yeasts was used (Illumina Inc., Berlin, Germany) to remove rRNA. Libraries were pooled and sequenced (paired-end, 2 x 75 cycles) on NextSeq 550 platform (Illumina Inc., Berlin, Germany). Three independent biological replicates were performed.
For RNA sequencing, the raw sequence files generated (.fastq files) underwent quality control analysis using the FastQC tool (http://www.bioinformatics.babraham.ac.uk/projects/fastqc), and the quality-checked paired-end reads were then aligned to the reference Candida albicans SC5314 genome (assembly GCA_000182965.3) using STAR (version 2.5.2a) (Dobin et al., 2013), with standard parameters. The FeatureCount algorithm (Liao et al., 2014) was used to quantify each transcript using the reads mapped to the genome. A given gene was considered expressed when detected by at least 10 total reads in the 3 replicates. Data normalization and differentially expressed transcripts were identified using DESeq2 (Love et al., 2014) with standard parameters; differential expression was reported as fold change. A gene with FDR ≤ 0.05 (False Discovery Rate) and with a value of Fold Change ≤ -1.5 (for down-regulated genes) or Fold Change ≥ 1.5 (for up-regulated genes) was considered significantly differentially expressed. For ChIP-sequencing, the analysis was performed using the Galaxy tool (v 22.05) (https://usegalaxy.eu/) (Galaxy Community, 2022). Briefly, after the FastQC quality check, the paired-end reads were aligned to the reference Candida albicans SC5314 genome (assembly GCA_000182965.3) using Bowtie 2 (Galaxy Version 2.4.4), and the generated BAM files were filtered with Filter BAM (-q=20) (Galaxy Version SAMTOOLS: 1.8). Mapped reads were indexed and merged using samtools MergeSamFiles (Galaxy Version 2.18.2.1) and converted to bigwig files using deepTools bamCoverage (Galaxy Version 3.5.1.0.0) with a bin size of 10 and normalization to genomic content. Peak calling was performed with MACS2 callpeak (Galaxy Version 2.2.7) using standard parameters for board regions and normalized to the effective genome size. Peak annotation was carried out using ChIPseeker (Galaxy Version 1.28.3). In both ChIP-seq and RNA-seq analyses, PCR duplicates were excluded. Network analysis was performed using ClueGo and CluePedia Cytoscape plugins (version 3.9.1) (Bindea et al., 2009). Venn diagrams were designed by InteractiVenn (Heberle et al., 2015).
Data are from at least three independent experiments, and results are expressed as means ± SD. Data were analyzed with GraphPad Prism 7 (GraphPad Software). Two-tailed Student’s t-test (2-group comparisons) or two-way ANOVA (>2-group comparisons) were performed as appropriate. P values < 0.05 were considered significant.
To determine the minimum concentration of NAM able to induce an accumulation of H3K56ac without affecting Candida growth within 28 h, different concentrations of NAM (5-100 mM) were assayed (data not shown). As shown in Figure 1A , 10 mM NAM did not significantly affect fungus duplication and, more importantly, caused a robust accumulation of H3K56 acetylation levels ( Figures 1B-D ). During C. albicans growth, nuclear protein fractions were prepared from selected time points (0, 2, 4, 6, 8, 10, and 28 h) from both treated and untreated cells to determine the acetylation levels of H3K56 by nanoscale Liquid Chromatography coupled to tandem Mass Spectrometry (nano-ESI-LC-MS/MS). The quantification of FQK(Ac)STELLIR, the histone tryptic peptide including the acetylated lysine, was performed by normalizing its chromatographic peak area versus YKPGTVALR, another tryptic peptide of H3 histone which does not display post-translational modifications. Nano-ESI-LC-MS/MS analysis revealed a maximum peak of H3K56 acetylation after 4 h-inoculation and a decrease over time in untreated control cells. On the contrary, the acetylation levels of H3K56 upon NAM treatment increased and accumulated during growth, reaching a plateau up to 24 h ( Figure 1B ). These results describe for the first time how H3K56 is acetylated during C. albicans yeast growth and confirm the inhibitory effect of NAM on the fungal sirtuin Hst3p. The NAM-depending accumulation of H3K56ac was also confirmed by Western blotting of histones isolated from Candida after 28 h-incubation ( Figures 1C, D ).
As a dimorphic fungus, C. albicans can reversibly switch from the yeast morphology to elongated hyphal forms, responding to environmental stimuli in a finely regulated process characterized by extensive changes in gene expression profile. Given the central role of the chromatin landscape in transcription activation/repression in eukaryotes and considering that H3K56 acetylation is the most abundant post-translational modification in C. albicans, we wondered whether such histone modification might be involved in the regulation of hypha-specific genes (HSG). To this end, we analyzed the colony morphology of C. albicans resulting from inhibition of Hst3p. The filamentation was induced with 10% serum, M199, or Spider medium at 37°C, in the presence or absence of 25 mM NAM. As reported in Figure 2 , 25 mM NAM led to a significant alteration in C. albicans morphology with a robust inhibition of filamentation and formation of hyphal crown around the macro-colonies on solid media. In addition, according to Wurtele and colleagues (Wurtele et al., 2010), we observed the formation of abnormal and enlarged filamentous structures, with a particular conformation called V-shaped hyphae, when Hst3p was inhibited under yeast-promoting conditions (YPD at 25°C) ( Figure 2 ). Such morphological alteration supports the hypothesis that H3K56ac might be involved in the yeast-to-hyphae transition. However, Western blotting analysis of histones isolated respectively from yeast, germ tube, and hyphal shape revealed that the overall acetylation levels of H3K56 are comparable among the three cellular forms ( Supplementary Figure S1 ).
As a member of a NAD+-dependent histone deacetylases family, Hst3p is inhibited, in a non-specific and non-selective way, by NAM. In order to verify if commercially available sirtuin inhibitors affect Hst3p activity, we tested Sirtinol, SirReal2, and Inauhzin, which are specific SIRT1 and SIRT2 inhibitors. In detail, yeast cells from an overnight culture were inoculated in YPD containing 10 µM Sirtinol, 50 µM SirReal2, or 50 µM Inauhzin (the highest concentrations at which the inhibitors are fully soluble), or with 10 mM NAM, here used as control. Yeast growth was followed up to 24 h by Time Lapse imaging. No morphological alterations were observed with any of the inhibitors used, whereas cells treated with NAM formed hyphae with V-shaped branches as expected ( Figure 3 ). Moreover, Western blotting of histones extracted from overnight treatments confirmed that Sirtinol, SirReal2, and Inauhzin did not induce a significant accumulation of H3K56ac, whereas NAM inhibited Hst3p as showed by the higher level of acetylated H3K56 ( Figure 4 ).
ChIP-seq is a powerful tool for the analysis and mapping of epigenetic marks. In order to identify the H3K56ac patterns across the Candida albicans genome, control and NAM-treated V-shaped cells were analyzed by ChIP-seq with anti-H3K56ac antibody after 28 h of incubation in YPD at 25°C. Plot profiles in Figure 5 show that H3K56ac is mainly localized in genomic regions across the TSS of genes. We identified 671 and 843 ChIP-enriched regions in CTRL and CaNAM, respectively ( Supplementary Dataset 1 ) ( Figure 6 ). Since histone acetylation is mainly associated with open chromatin and consequent transcriptional activation, we focused on ChIP-enriched promoter regions (2 kb upstream TSS). Among them, 283 enriched regions are common to both experimental conditions and are mainly in the promoters of essential genes involved in metabolic process, transcriptional and translational control (i.e., transcription initiators, ribosomal proteins, translation elongator factors) ( Supplementary Figure S2 ). Regions enriched for this histone modification only upon NAM treatment were 447, indicating that in those regions, H3K56 is likely deacetylated by Hst3p in control cells ( Figure 6 ). To better understand the biological significance of differentially acetylated regions, we performed a functional analysis using ClueGo and CluePedia Cytoscape plugins, which integrates Gene Ontology (GO) terms to create a functionally organized GO term network, and we found significant enrichment for 8 GO biological processes (Bonferroni adjusted pValue ≤0.05). These differentially acetylated regions include promoters of genes involved in filamentation, phenotypic switching, and adhesion which would explain the abnormal V-shaped morphologies observed in CaNAM ( Figure 7 ). Among them, we found several transcription factors involved in morphological processes: OFI1, involved in the regulation of white-opaque switching and filamentous growth (Du et al., 2015); WAL1, involved in polarized hyphal growth (Walther and Wendland, 2004); EFG1, a central transcriptional regulator of morphogenesis and biofilm (Glazier, 2022); NRG1, a transcription factor/repressor regulating hyphal gene and virulence (Braun et al., 2001); ACE2, which regulates morphogenesis, adherence, and virulence (Kelly et al., 2004); WOR1, the master regulator of white-opaque phenotypic switching (Huang et al., 2006), WOR2, WOR3 and WOR4 which, together with WOR1, CZF1, EFG1, and AHR1, form the transcriptional network that triggers the ability to switch between white and opaque cell types (Hernday et al., 2013; Lohse and Johnson, 2016). In addition, we found genes involved in adhesion and biofilm formation, such as HWP1, a hypha-specific cell surface protein required for biofilm formation in vivo (Nobile et al., 2006), and HWP2, a GPI-anchored protein required for biofilm formation, adhesion, filamentous growth (Hayek et al., 2010), the transcription factor CRZ2, required for pH-induced filamentation (Kullas et al., 2007), the adhesins ALS3, a fungal invasin that mimics host cell cadherins and induces endocytosis by binding to N-cadherin on endothelial cells and E-cadherin on oral epithelial cells (Phan et al., 2007).
To integrate and strengthen ChIP-seq data and to better understand how H3K56 acetylation levels influence the C. albicans transcriptome and, consequently, the V-shaped hyphae development, we performed RNA sequencing of CaNAM vs. CTRL cells. We found 1330 upregulated (FC ≥ 1.5; FDR ≤ 0.05) and 1081 downregulated (FC ≤ -1.5; FDR ≤ 0.05) transcripts in CaNAM compared to CTRL cells ( Supplementary Figures S3 and 4 ). Consistent with V-shaped morphology, Gene Ontology analysis revealed that the upregulated genes are mainly related to white-opaque, filamentation, cell wall organization, and adhesion ( Figure 8 ). Intersecting RNA sequencing data with the 447 ChIP-enriched regions of NAM-treated samples, we found 87 genes whose up-regulation directly correlates with H3K56 acetylation patterns associated with their promoters, including OFI1, WOR1, WOR2, HWP1, NRG1, ACE2, CRZ2, the secreted aspartyl proteinases SAP1 and SAP2, aside from some GPI-anchored adhesin-like protein known to be involved in adhesion and virulence ( Supplementary Dataset 2 ). Unexpectedly, a subset of 85 genes whose promoters displayed H3K56 acetylation only upon NAM inhibition showed an opposite trend, with higher acetylation but lower transcript abundance in CaNAM compared to CTRL, suggesting that H3K56 acetylation is not always followed by a higher transcription rate. This could be explained by the fact that in all eukaryotic organisms, a wide range of mechanisms contribute to transcription control (including mRNA stability control, Pol II termination, and Pol II attenuation) and is consistent with NET-seq analysis carried out by Topal and colleagues (Topal et al., 2019). They showed that, besides the leading role in enhancing transcription initiation, H3K56ac might transiently function as a transcriptional repressor by promoting nucleosome assembly in S. cerevisiae. Furthermore, our ChIP-seq analysis revealed that Hst3p also regulates H3K56ac across the promoters of genes encoding the RNA polymerase II mediator complex subunits MED1, the RNA polymerase II regulator EED1, besides other predicted ribonucleases or putative proteins with a predicted role in mRNA 3’-end processing, mRNA polyadenylation, and pre-mRNA cleavage. These results suggest a possible role of H3K56ac in the modulation of a fine-tuning mechanism of transcriptional control. In addition to the 87 genes whose expression seems directly related to H3K56ac associated with their promoters, our transcriptome analysis showed a subset of genes whose expression is regulated indirectly by such histone mark. For instance, our ChIP-seq revealed that Hst3p inhibition results in H3K56 acetylation associated with the promoter of EFG1, a key transcription regulator of several morphogenesis and virulence-related genes, whose transcription is not only negatively autoregulated by Efg1, but is also repressed by a series of positive feedback loops established by Wor1, Wor2, and Czf1 (Hernday et al., 2013). Moreover, ECE1, which encodes for the cytolytic peptide candidalysin, and UME6, the filament-specific transcriptional regulator, are up-regulated in RNA-seq upon NAM treatment and directly regulated by EFG1 (Banerjee et al., 2008; Moyes et al., 2016). Together our results reveal a complex mechanism behind the gene regulation mediated by H3K56ac, underlining the importance of such histone mark in controlling a wide range of biological processes in C. albicans.
Nosocomial infections caused by Candida spp. arise great concern due to the emergence of resistant fungal strains to the main antifungal drugs. In particular, C. albicans is the most prevalent species causing disease in both adult and pediatric populations, with a 30-day mortality rate approaching 40% in immunocompromised hosts (Koehler et al., 2019). The similarity between fungal and human cells limits the identification of specific targets that uniquely affect the microbial cell without damaging the host, further reducing the therapeutic options for treating fungal infections (Salazar et al., 2020). In addition, the intrinsic resistance to drugs such as azoles, echinocandins, or polyenes (Nami et al., 2019) of some pathogenic Candida species, brings the further need to develop antifungals with novel mechanisms for medical practice (Caruso et al., 2020). In C. albicans, the most abundant post-translational modification is the H3K56 acetylation, regulated by the histone acetyltransferase Rtt109p and the histone deacetylase Hst3p. Histone acetylation/deacetylation balance plays a crucial role in C. albicans growth and virulence (Su et al., 2020). Interestingly, Hst3p displays sequence motifs not shared with human sirtuins (Wurtele et al., 2010); therefore, this histone deacetylase represents a promising therapeutic target for developing new antifungal agents. Several known HDAC inhibitors have already been explored as new potential antifungal therapy, but the selectivity of such molecules remains the main problem (Su et al., 2020). Indeed, we tested three currently available sirtuin inhibitors (Inauhzin, Sirtinol, and SirReal2), but none showed activity on Hst3p since they did not induce either morphological alterations in Candida or a significant accumulation of H3K56ac. Therefore, we enrolled NAM as a non-specific sirtuin inhibitor to explore the effects of H3K56ac accumulation in Candida albicans. First, we examined the trend of H3K56 acetylation levels during the Candida yeast growth curve. In particular, by nano-ESI-LC-MS/MS, we observed that H3K56 acetylation reaches a maximum at the beginning of the logarithmic phase and then decreases, consistent with the observation that the H3K56ac marks newly synthesized histones, facilitating their deposition onto chromatin. On the contrary, when Hst3p is inhibited by NAM during yeast growth, we observed an accumulation of H3K56 acetylation over time, reaching a plateau at 24 h. Moreover, Hst3p inhibition results in abnormal phenotypes; in particular, when Candida is grown under hyphae-inducing conditions, there is a substantial filamentation reduction. In contrast, as previously observed by Wurtele and colleagues (Wurtele et al., 2010), when Hst3p is inhibited under yeast-promoting conditions, Candida forms an abnormal phenotype named V-shaped hyphae. However, we observed that the overall acetylation levels of H3K56 do not change significantly between yeasts, germ tubes, and hyphae, suggesting that this histone modification, weakening inter-nucleosomal interactions and leading to the chromatin relaxation for transcription (Wang and Hayes, 2008), is necessary for all morphogenetic phases of Candida. Since the V-shaped hyphae is a peculiar morphology associated with increased levels of H3K56ac, we used these cells as a model to identify Hst3p targets. Although H3K56 acetylation is widespread in Candida, our ChIP-seq analysis revealed a strong enrichment across the TSS of genes related to Candida sp. pathogenicity, and morphology. Noteworthy, we identified 447 regions ChIP enriched only upon NAM treatment, including genes encoding for HSG, adhesin proteins, degradative enzymes, and white-opaque switching, such as OFI1, ACE2, WOR1, WOR2, WOR3, WOR4, HWP1, HWP2, CRZ2, ALS3. Moreover, our RNA-seq analysis showed a significant transcriptome dysregulation upon Hst3p inhibition, with 1330 up-regulated and 1081 down-regulated transcripts in 28 h CaNAM compared to CTRL. Gene Ontology analysis confirmed that, among the up-regulated transcripts, there are genes mainly related to filamentation, cell wall organization, and adhesion giving rise to the possibility that H3K56, through its acetylation status, regulates these processes. Interestingly, we identified 87 genes whose transcriptional increase well correlates with the enrichment of H3K56 acetylation on their promoters, including some well-known regulators of phenotypic switching and virulence. For instance, the zinc-finger transcription factor OFI1, whose over-expression promotes filamentous growth in several culture conditions (Du et al., 2015) and WAL1, required for the organization of the cortical actin cytoskeleton and for the polarized hyphal growth (Walther and Wend land, 2004). Moreover, our ChIP-seq revealed the presence of H3K56ac patterns in the promoter of EFG1, whose transcript abundance is lower in CaNAM compared to CTRL. This is consistent with previous studies showing that, overexpression of EFG1 facilitates hyphal initiation but, shortly thereafter, leads to Efg1 dependent transcript down-regulation (Tebarth et al., 2003; Lassak et al., 2011). Furthermore, EFG1 is repressed both directly and indirectly by Wor1 (Hernday et al., 2013) that is up-regulated upon NAM treatment. Efg1, acting via cAMP-PKA pathway, regulates the expression of several genes involved in the filamentation, such as UME6, a known regulator of hyphal extension, upregulated in the V-shaped conditions (Banerjee et al., 2008). These results might explain the abnormal phenotype, namely V-shaped hyphae, observed in C. albicans cells exposed to NAM in yeast promoting conditions. We are aware that the use of a non-specific inhibitor may not be the best way to proceed, but several indications suggest that the effects of NAM reported in this study are exerted mainly through inhibition of Hst3p activity. Firstly, in C. albicans, the deacetylation of H3K56 occurs via Hst3p, thus the accumulation of H3K56ac following NAM treatment is likely due to the inhibition of Hst3p. In addition, in our experimental conditions, we used a concentration of NAM that does not affect Candida growth, but induces the formation of V-shaped hyphae, the same phenotype observed by Wurtele and coworkers in the heterozygous deletion mutant of HST3. Moreover, although in C. albicans, besides Hst3p there are other sirtuins, such as Sir2p, Hst1p, and Hst2p, Zhao and Rusche (2021) demonstrated that true hypha formation was significantly reduced only by the deletion of SIR2 but not of HST1 or HST2. Moreover, the expression of hypha-specific genes HWP1, ALS3, and ECE1 was significantly reduced in a sir2 single mutant compared to the wild type. In contrast, in our experimental conditions, all three of these genes are up-regulated, suggesting that the inhibition of Hst3p obtained with NAM predominates over that eventually exerted on Sir2p. Overall, our study represents the first genome-wide H3K56 acetylation analysis in C. albicans and provides the first map of H3K56ac patterns across the C. albicans genome, representing a rich resource for future studies. Nevertheless, the evidence that H3K56ac directly regulates the expression of several virulence-related genes points out the relevance of such epigenetic modification in regulating C. albicans virulence, confirming that Hst3p is an appealing target for the development of new potential antifungal drugs.
The datasets presented in this study can be found in online repositories. The name of the repository and accession number(s) can be found below: ArrayExpress, accession E-MTAB-12193; E-MTAB-12167.
MC: data curation, investigation, methodology, visualization, bioinformatic analysis, writing—original draft preparation, and writing—review and editing; DE: data curation, investigation, methodology, visualization, and writing—review and editing; MP: investigation and methodology; AMP: investigation and methodology; MM: methodology and writing—review and editing; CC: methodology; GG: methodology and bioinformatic analysis; FR: methodology and bioinformatic analysis; PT: methodology and writing—review and editing, AnP: data curation and investigation; AT: conceptualization, data curation, investigation, validation, writing—review and editing, and funding acquisition; AmP: conceptualization, data curation, investigation, validation, visualization, writing—original draft, writing—review and editing, and funding acquisition. All authors contributed to the article and approved the submitted version.
This work was supported by University of Salerno intramural funds FARB.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647177 | Sung-Hee Yoon,Geun-Bae Kim | Inhibition of Listeria monocytogenes in Fresh Cheese Using a Bacteriocin-Producing Lactococcus lactis CAU2013 Strain | 01-11-2022 | Lactococcus lactis,bacteriocin,Listeria monocytogenes,cheese starter culture,foodborne pathogen | Abstract In recent years, biocontrol of foodborne pathogens has become a concern in the food industry, owing to safety issues. Listeria monocytogenes is one of the foodborne pathogens that causes listeriosis. The major concern in the control of L. monocytogenes is its viability as it can survive in a wide range of environments. The purpose of this study was to isolate lactic acid bacteria with antimicrobial activity, evaluate their applicability as a cheese starter, and evaluate their inhibitory effects on L. monocytogenes. Lactococcus lactis strain with antibacterial activity was isolated from raw milk. The isolated strain was a low acidifier, making it a suitable candidate as an adjunct starter culture. The commercial starter culture TCC-3 was used as a primary starter in this study. Fresh cheese was produced using TCC-3 and L. lactis CAU2013 at a laboratory scale. Growth of L. monocytogenes (5 Log CFU/g) in the cheese inoculated with it was monitored during the storage at 4°C and 10°C for 5 days. The count of L. monocytogenes was 1 Log unit lower in the cheese produced using the lactic acid bacteria strain compared to that in the cheese produced using the commercial starter. The use of bacteriocin-producing lactic acid bacteria as a starter culture efficiently inhibited the growth of L. monocytogenes. Therefore, L. lactis can be used as a protective adjunct starter culture for cheese production and can improve the safety of the product leading to an increase in its shelf-life. | Inhibition of Listeria monocytogenes in Fresh Cheese Using a Bacteriocin-Producing Lactococcus lactis CAU2013 Strain
In recent years, biocontrol of foodborne pathogens has become a concern in the food industry, owing to safety issues. Listeria monocytogenes is one of the foodborne pathogens that causes listeriosis. The major concern in the control of L. monocytogenes is its viability as it can survive in a wide range of environments. The purpose of this study was to isolate lactic acid bacteria with antimicrobial activity, evaluate their applicability as a cheese starter, and evaluate their inhibitory effects on L. monocytogenes. Lactococcus lactis strain with antibacterial activity was isolated from raw milk. The isolated strain was a low acidifier, making it a suitable candidate as an adjunct starter culture. The commercial starter culture TCC-3 was used as a primary starter in this study. Fresh cheese was produced using TCC-3 and L. lactis CAU2013 at a laboratory scale. Growth of L. monocytogenes (5 Log CFU/g) in the cheese inoculated with it was monitored during the storage at 4°C and 10°C for 5 days. The count of L. monocytogenes was 1 Log unit lower in the cheese produced using the lactic acid bacteria strain compared to that in the cheese produced using the commercial starter. The use of bacteriocin-producing lactic acid bacteria as a starter culture efficiently inhibited the growth of L. monocytogenes. Therefore, L. lactis can be used as a protective adjunct starter culture for cheese production and can improve the safety of the product leading to an increase in its shelf-life.
Listeriosis is a foodborne disease caused by Listeria monocytogenes. It can lead to sepsis, meningitis, encephalitis, and even death (de Noordhout et al., 2014). Despite its low incidence compared with that of other foodborne illnesses, listeriosis is one of the major issues in the food industry because of its high fatality rate. L. monocytogenes is found in dairy products, particularly in ready-to-eat cheese products. As L. monocytogenes survives in various environments such as a those with a wide range of temperature (0°C–45°C) and pH (4.1–9.6), it can contaminate cheese at several stages of production; therefore, its growth is difficult to control (Lungu et al., 2009; Melo et al., 2015). Several methods have been used to control the growth of L. monocytogenes in cheese, including using bacteriocin or bacteriocin-producing lactic acid bacteria (LAB). Bacteriocins are peptides or proteins, ribosomally synthesized by bacteria, which have antimicrobial ability against closely related species. The application of bacteriocin-producing bacteria is advantageous as they are stable, cost-effective, and safe. Anti-listerial activity of LAB in cheese have also been reported (Coelho et al., 2014; Dal Bello et al., 2012; Kondrotiene et al., 2018). In the present study, we aimed to determine the effects of bacteriocin-producing LAB isolated from raw milk on the growth of L. monocytogenes in milk broth and cheese.
Potential bacteriocin-producing LAB were isolated from raw bovine milk, obtained from a Chung-Ang University-affiliated farm (Anseong, Korea). The sample was serially diluted ten-fold and plated on MRS agar (BD Difco, Franklin Lakes, NJ, USA). The plates were incubated at 37°C for 24–48 h, and a total of 90 well-isolated colonies were collected. Each colony was inoculated into MRS broth for 24 h at 37°C. To screen for antimicrobial activity, the cell-free supernatant (CFS) was obtained after neutralization with 1N NaOH, centrifugation at 15,000×g for 10 min at 4°C, and filtered through 0.45 μm filters to remove bacterial cells. Then, each supernatant was spotted on the tryptic soy agar (TSA; BD Difco) plate inoculated with a lawn of L. monocytogenes ATCC 19115 as an indicator strain. The plates were incubated at 30°C for 12 h, and antibacterial activity was confirmed with the presence of inhibition zone. The strains with antibacterial activity were routinely cultured in MRS broth at 37°C overnight and were preserved in 10% skim milk supplemented with 25% (v/v) glycerol, stored at –80°C for further use.
The bacteriocin-producing strains were identified by Gram staining, carbohydrate fermentation profile [analytical profile index (API) test], and 16S rRNA gene sequencing analysis. Gram staining and API analysis was performed using a Gram-stain kit (BD Difco) and API 50 CHL kit (bioMérieux, Marcy-l’Étoile, France), respectively, according to the manufacturer’s instructions. For 16S rRNA analysis, the genomic DNA was extracted using QIAamp PowerFecal DNA Kit (Qiagen, Hilden, Germany) and amplified using 2X H-star Taq polymerase chain reaction (PCR) Master Mix (BioFACT, Daejeon, Korea). PCR was performed using the universal bacterial primers 27F (5′-AGAGTTTGATCMTGG CTCAG-3′), 1492R (5′-TACGGYTACC TTGTTACGACTT-3′), 785F (5′-GGATTAGA TACCCTGGTA-3′), and 805R (5′-GACTACCAGGGTATCTAATC-3′). The PCR products were purified using a PCR purification kit (Qiagen) and sequenced by SolGent (Daejon, Korea). The analyzed sequences were confirmed using the EzTaxon-e server (https://www.ezbiocloud.net/; Kim et al., 2012) and NCBI GenBank database using the Basic Local Alignment Search Tool (BLAST) algorithm (https://blast.ncbi.nlm.nih.gov/Blast.cgi; Altschul et al., 1990).
Bacteriocin activity was assessed using a spot-on-lawn method as described previously (Phumisantiphong et al., 2017) with minor modifications. Briefly, each indicator strain was inoculated to 4 mL of molten TSA and overlaid on the base TSA plate. After solidification, 20 μL of the neutralized CFS of LAB strains was spotted onto the indicator lawn. After incubation at 30°C for 12 h, a clear inhibition zone was observed. The foodborne pathogens used as indicator strains were cultured in tryptic soy broth (TSB; BD Difco) at 37°C overnight before use. The experiment was conducted in triplicates.
The acid production of the Lactococcus lactis CAU2013 was evaluated and compared with that of the commercial starter TCC-3 (Chr. Hansen, Hørsholm, Denmark), which consisted of Lactobacillus delbrueckii subsp. bulgaricus and Streptococcus thermophilus. The cultures were grown in MRS broth at 37°C overnight. Then, individual cultures and mixture of L. lactis CAU2013 and TCC-3 (1:1 ratio) were inoculated in 10% skim milk broth and whole milk. The pH and titratable acidity (TA) were measured every three hours for 12 h while incubating at 30°C. To determine TA, 0.1% phenolphthalein was used as an indicator, and 0.1 N sodium hydroxide (NaOH) for titration.
To determine the anti-listerial properties of strain CAU2013 when used as an adjunct starter in milk, 10% skim milk broth and whole milk media were inoculated with an overnight culture of CAU2013 and 1:1 ratio of CAU2013 and TCC-3 starter (final concentration of 7 Log CFU/mL). Additionally, the overnight culture of L. monocytogenes ATCC 19115 was inoculated to each setup (final concentration of 5 Log CFU/mL). The inoculated milk media were incubated at 30°C for 12 h. The viable cell count of L. monocytogenes was determined every three hours. Samples were diluted serially in ten-fold increments using 1×phosphate-buffered saline (PBS; pH 7.5) and plated on Oxford agar (BD Difco).
The lab-scale cheese was manufactured following the methods of Mills et al. (2011) with some modifications. TCC-3 was used as the primary starter and L. lactis CAU2013 as an adjunct culture. The starter cultures were initially grown in MRS broth at 37°C for 24 h before inoculation into 10% skim milk broth and incubated for 18 h at 37°C before use. Additionally, L. monocytogenes ATCC19115 was cultured in TSB for 18 h at 37°C before use. Milk (400 mL; Seoul Milk, Seoul, Korea) was heated to 31°C before the inoculation of starter culture. The starter cultures were inoculated as follows: TCC-3 and L. lactis CAU2013 and TCC-3 (1:1 ratio), both at a final concentration of 7 Log CFU/mL. Subsequently, 0.01% L. monocytogenes at a level of 5 Log CFU/mL was inoculated into both treatments. After 30 min, 0.2 g/L of rennet was added, and the mixture was stirred for 2 min. Once coagulum formed firmly, the curd was cut into cubes, and the mixture was stirred for 10 min. Then, the mixture was heated to 36°C for 10 min and stirred for 20 min. The whey was drained off, and curd was distributed into the sterile dish. The samples were stored at 4°C and 10°C for 5 days. The procedure of cheese production is illustrated in Fig. 1.
The viable cell counts of LAB and L. monocytogenes in the lab-produced cheese were determined in duplicate every day during storage at 4°C and 10°C. For microbial analysis, 1 g of cheese was homogenized in 9 mL of PBS buffer and were serially diluted ten-fold in the same buffer and plated on the appropriate agar plate. The LABs were enumerated on MRS agar after incubation at 37°C for 3 days, and L. monocytogenes on Oxford agar after incubation at 37°C for 24 h. All of the experiments were conducted in triplicates.
Among the 90 colonies isolated from raw milk, one isolate exhibited antibacterial activity against L. monocytogenes. The strain CAU2013 was characterized as a gram-positive, coccus-shaped bacterium. The biochemical characteristics determined using the API 50 CHL kit are described in Table 1. 16S rRNA gene sequence analysis revealed that strain CAU2013 is most likely a strain of L. lactis (Table 2), which commonly produce nisin (Shin et al., 2016). Neighbor-joining phylogenetic tree of the strain CAU 2013 and related type strains based on 16S rRNA gene sequences also clearly show that this strain belongs to L. lactis (Fig. 2). L. lactis strains are historically used in the fermentation and preservation of food and are generally recognized as safe (Cook et al., 2018). Therefore, L. lactis CAU2013 was selected for downstream applications in the study. L. monocytogenes ATCC 19115 was used as an indicator strain for all experiments because it belongs to the serotype 4b, which causes most cases of listeriosis.
The bacteriocin produced by L. lactis CAU2013 had antibacterial activity against all Listeria strains as well as Staphylococcus aureus, which are common foodborne pathogens (Yoon, 2020). However, no antibacterial activity was observed against other gram-positive foodborne pathogens, such as Salmonella enteritidis and Escherichia coli (Table 3). Generally, nisin is highly effective against gram-positive bacteria by binding to lipid °C, which leads to the inhibition of cell wall biosynthesis or pore formation in the membrane. However, nisin cannot bind to its target lipid II in gram-negative bacteria, because of the presence of the outer membrane (Li et al., 2018).
The changes in pH and TA values in 10% skim milk broth and in whole milk are presented in Fig. 3. L. lactis CAU2013 reduced the pH of skim milk broth from 6.41 to 5.77 and that of whole milk from 6.65 to 6.20. Additionally, TA value increased to 0.25 in both broths. Ayad et al. (2004) described fast, medium, or slow-acidifying strains as ΔpH (=pHat time–pHzero time) of 0.4 U achieved after 3 h, 3–5 h, and >5 h, respectively. Also, Raquib et al. (2003) classified strains with TA as low, moderate, or fast when the TA values were <0.5, between 0.5 and 0.6, and >0.6, respectively. Therefore, L. lactis CAU2013 can be classified as a low acidifier strain. This result is consistent with other studies that reported poor acid production from L. lactis strains (Ayad et al., 2004; Coelho et al., 2014). The pH values measured corresponded with the calculated TA and were generally similar for skim milk broth and whole milk. The mixed starter, consisting of TCC-3 and L. lactis CAU2013, accelerated the acidification in milk. Nevertheless, bacteriocin-producing strains delay acidification (Garde et al., 1997); however, the strain CAU2013 did not show similar properties. The accelerated acidification might be because of the interaction between the strains; however, the underlying mechanisms need further research. Ávila et al. (2005) observed that enterocin-producing adjunct starter enterococci enhanced milk acidification, which may be stimulated by the low-molecular-weight nitrogen compounds produced by primary starter, Lactobacillus helveticus LH92. The rapid decline in pH during the initial stage of cheese production is crucial for curd formation and prevention of the growth of undesirable microorganisms. Therefore, the fast-acidifying strains can be used as primary starters, while the slow-acidifying bacteria can be used as adjunct starters. As the strain CAU2013 has antibacterial property but has low acid production ability, it is better suited as an adjunct starter culture.
The growth of L. monocytogenes was monitored in skim milk broth and whole milk during incubation at 30°C. In skim milk broth with L. lactis CAU2013, the concentration of L. monocytogenes count was reduced by 3 Log units more compared with that of other samples after 3 h and not detected following 6 h of fermentation (Fig. 4A). In the whole milk with the strain CAU2013, L. monocytogenes count was reduced by 0.5 Log unit after 6 h, and 1 Log unit after 9 h compared with that of other samples (Fig. 4B). The results support the findings from several studies that reported that the addition of bacteriocin affects the biocontrol of spoilage bacteria. Muñoz et al. (2007) investigated that E. faecalis-produced enterocin in milk and found that it could control the growth of Staphylococcus aureus. In addition, according to Arqués et al. (2011), the addition of nisin in milk decreased L. monocytogenes count by 3 Log units after 4 h. The efficiency of the combined starter cultures in the inhibition of L. monocytogenes was lower in whole milk than in skim milk. The difference in the composition between the two milk media could be a factor responsible for the difference. In addition, Muñoz et al. (2007) stated that low effectiveness in foods could be attributed to higher retention of the bacteriocin molecules by milk components, resulting in slower diffusion. However, in both cases, inhibition of L. monocytogenes growth was observed. The results suggest the potential application of L. lactis CAU2013 in various food systems to control L. monocytogenes growth.
The cell count of the starter cultures was determined during the storage at 4°C and 10°C (Fig. 5). In both the cases, LAB reached a final concentration of 9 Log CFU/g during cheese manufacture. During the storage at 4°C, the cheese treated with TCC-3 starter culture maintained L. monocytogenes count at 7.5 to 7.7 Log CFU/g. In contrast, the cheese treated with TCC-3 and L. lactis CAU2013 had less L. monocytogenes count, approximately 0.5 Log unit at 0 h and 1 Log unit after 5 days with a final concentration of 6.4 Log CFU/g. Besides, during the storage at 10°C, the cheese treated with TCC-3 starter culture maintained the bacterial count between 6.86 and 7.31 Log CFU/g (Fig. 5A). within contrast, the cheese treated with TCC-3 and CAU2013 had less L. monocytogenes count, approximately 1 Log unit at 0 h and 1.5 Log unit after 5 days, with a final concentration of 5.76 Log CFU/g (Fig. 5B). This result is consistent with a study that reported that 2 Log unit reduction was observed in cheese with L. lactis strain (Coelho et al., 2014). Moreover, Kondrotiene et al. (2018) showed that nisin-producing L. lactis strains decreased the growth of L. monocytogenes in fresh cheese during 7 days of storage at 4°C. Therefore, the results support that manufacturing cheese using a bacteriocin-producing starter reinforced the inhibition of growth of L. monocytogenes, and it would be effective in controlling contamination during cheese production. Additionally, after storage at temperatures of 4°C and 10°C, L. monocytogenes count was reduced, which may confirm the potential of LAB in controlling the growth of L. monocytogenes during storage at refrigeration temperature. |
|
PMC9647179 | Biao Xuan,Jongbin Park,Geun-Shik Lee,Eun Bae Kim | Oral Administration of Mice with Cell Extracts of Recombinant Lactococcus lactis IL1403 Expressing Mouse Receptor Activator of NF-kB Ligand (RANKL) | 01-11-2022 | cell extracts,receptor activator of NF-kB ligand (RANKL),Lactococcus lactis,transcriptome,microbiome | Abstract Receptor activator of NF-kB ligand (RANKL) is known to play a major role in bone metabolism and the immune system, and its recombinant form has been expressed in bacterial systems for research since the last two decades. However, most of these recombinant forms are used after purification or directly using living cells. Here, there were cell extracts of recombinant Lactococcus lactis expressing mouse RANKL (mRANKL) used to evaluate its biological activity in mice. Mice were divided into three groups that were fed phosphate-buffered saline (PBS), wild-type L. lactis IL1403 (WT_CE), and recombinant L. lactis expressing mRANKL (mRANKL_CE). The small intestinal transcriptome and fecal microbiome were then profiled. The biological activity of mRANKL_CE was confirmed by studying RANK-RANKL signaling in vitro and in vivo. For small intestinal transcriptome, differentially expressed genes (DEGs) were identified in the mRANKL_CE group, and no DEGs were found in the WT_CE group. In the PBS vs. mRANKL_CE gene enrichment analysis, upregulated genes were enriched for heat shock protein binding, regulation of bone resorption, and calcium ion binding. In the gut microbiome analysis, there were no critical changes among the three groups. However, Lactobacillus and Sphingomonas were more abundant in the mRANKL_CE group than in the other two groups. Our results indicate that cell extracts of mRANKL_CE can play an effective role without a significant impact on the intestine. This strategy may be useful for the development of protein drugs. | Oral Administration of Mice with Cell Extracts of Recombinant Lactococcus lactis IL1403 Expressing Mouse Receptor Activator of NF-kB Ligand (RANKL)
Receptor activator of NF-kB ligand (RANKL) is known to play a major role in bone metabolism and the immune system, and its recombinant form has been expressed in bacterial systems for research since the last two decades. However, most of these recombinant forms are used after purification or directly using living cells. Here, there were cell extracts of recombinant Lactococcus lactis expressing mouse RANKL (mRANKL) used to evaluate its biological activity in mice. Mice were divided into three groups that were fed phosphate-buffered saline (PBS), wild-type L. lactis IL1403 (WT_CE), and recombinant L. lactis expressing mRANKL (mRANKL_CE). The small intestinal transcriptome and fecal microbiome were then profiled. The biological activity of mRANKL_CE was confirmed by studying RANK-RANKL signaling in vitro and in vivo. For small intestinal transcriptome, differentially expressed genes (DEGs) were identified in the mRANKL_CE group, and no DEGs were found in the WT_CE group. In the PBS vs. mRANKL_CE gene enrichment analysis, upregulated genes were enriched for heat shock protein binding, regulation of bone resorption, and calcium ion binding. In the gut microbiome analysis, there were no critical changes among the three groups. However, Lactobacillus and Sphingomonas were more abundant in the mRANKL_CE group than in the other two groups. Our results indicate that cell extracts of mRANKL_CE can play an effective role without a significant impact on the intestine. This strategy may be useful for the development of protein drugs.
The production of recombinant proteins has been the foundation of the industrial and pharmaceutical biotechnology for the past 30 years (Puetz and Wurm, 2019). There are a wide variety of cell factories, such as mammalian cell lines, insect cells, whole plants, yeast, and bacteria (Ferrer-Miralles and Villaverde, 2013). Lactococcus lactis subsp. lactis IL1403 is a laboratory strain that is widely used in recombinant DNA technology (Pedersen et al., 2005). L. lactis is generally recognized as safe by the Food and Drug Administration (FDA; Song et al., 2017). Moreover, L. lactis is a popular microbial factory system owing to the wealth of genetic knowledge about it, and several existing recombinant protein expression systems (Linares et al., 2010). With the emergence of safety issues related to live microbial cells, including living modified organisms (LMOs), there is increasing interest in probiotic cell components and metabolites (Teame et al., 2020). To date, there have been few studies on the oral administration of bacterial cell extracts as delivery vectors containing recombinant proteins. Microfold (M) cells are specialized immune cells located in gut-associated lymphoid tissue (GALT), and play an essential role in the initiation of the intestinal immune response that transports luminal antigens through the intestine toward GALT (Foussat et al., 2001). Many previous studies have shown that the receptor activator of NF-kB ligand (RANKL) can induce the differentiation of M cells (Kanaya et al., 2012; Knoop et al., 2009; Kunisawa et al., 2008). Knoop et al. (2009) showed that purified recombinant mouse RANKL produced by a recombinant Escherichia coli differentiated M cells in the small intestine of RANKL-null mice (Knoop et al., 2009). In addition, Kim et al. (2015) showed that oral administration of recombinant L. lactis secreting mouse RANKL significantly increased the number of mature M cells in the mouse small intestine (Kim et al., 2015). These studies demonstrated that recombinant RANKL produced by bacteria is biologically active. With the advent of next-generation sequencing (NGS) technology, it is easy to determine gene expression levels in the tissues or gut microbiomes. Many studies have focused on the profiling gene expression in diseases, and also profiled the alteration of gut microbiome in diseases or determined the effect of consuming live probiotics, but the alteration of gene expression and gut microbiome after intake of bacterial cell extracts or recombinant proteins is not clear yet. In this study, the effect of the cell extracts of recombinant L. lactis expressing mRANKL (mRANKL_CE) were evaluated on small intestinal gene expression and gut microbiome in mice, to determine the usefulness of combinations of cell extracts and recombinant proteins for the development of protein drugs.
Wild-type L. lactis IL1403 was used as the host strain for recombinant protein production, and grown in M17 medium (MBcell, Seoul, Korea) supplemented with 5 g/L of glucose (M17G). Recombinant L. lactis IL1403 was grown in M17G media with chloramphenicol (5 μg/mL) and erythromycin (5 μg/mL) at 30°C.
Mouse RANKL sequence was used in Knoop and Kim’s studies (Kim et al., 2015; Knoop et al., 2009). To secrete the target protein, the signal peptide of USP45 (van Asseldonk et al., 1993) was added to the N-terminus, and to detect the expressed protein, a his-tag (his6x) was added to the C-terminus of the target gene. The designed amino acid sequence was codon-optimized using DNAWorks v3.2.4 (Hoover and Lubkowski, 2002) based on the L. lactis IL1403 codon usage table. The insert fragment was synthesized by Macrogen (Seoul, Korea). The insert fragment was shown in Supplementary Fig. S1. The plasmid vector pILPtuf.Mb was used as the backbone (Kim et al., 2009). Insert and vector were ligated at the NdeI and XhoI restriction sites and transformed into wild-type L. lactis IL1403 (WT_CE) competent cells. Vector construction is shown in Fig. 1A.
Wild-type and recombinant L. lactis strains were cultured in 50 mL of M17G broth without and with antibiotics, respectively. The optical density of each sample was measured at a wavelength of 600 nm every 1 h for 12 h and 24 h after inoculation.
The expression of mRANKL from recombinant L. lactis was measured by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and western blotting. Wild type and recombinant L. lactis were cultured in M17G broth without and with antibiotics at 30°C for 10 h, respectively. For preparation of cell extracts, 10 mL of cultured cells were harvested by centrifugation at 12,300×g for 1 min and lysed using a bead beater with 0.5 g sterilized glass beads (0.5 mm) and 200 μL 1×phosphate buffered saline (PBS). The secreted proteins were collected from 10 mL of cultured cell-free supernatant (0.2 μm filtered) through trichloroacetic acid precipitation, and dissolved in 200 μL 1×PBS for SDS-PAGE and western blot analysis. To quantify the amount of mRANKL produced, 18 kDa commercial recombinant His-tagged human calmodulin (Merck KGaA, Darmstadt, Germany) was used for construction of the standard curve using amounts of 1.5, 1, and 0.5 μg.
RAW 264.7 cells were seeded on cell culture dishes in Dulbecco’s modified Eagle’s medium (DMEM; Capricorn, Düsseldorf, Germany) with 10% fetal bovine serum (FBS; Capricorn) and 1% penicillin and streptomycin (P/S) at a density of 2×106 cells/mL. After seeding for 4 h, the medium was changed to DMEM with 10% FBS, 1% P/S, and 30 ng/mL macrophage colony-stimulating factor with crude cell extracts of wild-type or recombinant L. lactis, that were prepared as described above. The group treated with 1×PBS as control was named PBS, and groups treated with cell extracts from wild type L. lactis IL1403 and mRANKL producing L. lactis were named WT_CE and mRANKL_CE, respectively. According several dose test, finally, the cell extracts containing 90 ng/mL of mRANKL used in this study, and the same amount of cell extracts of WT_CE were used. Commercial mouse RANKL (Abcam, Hanam, Korea) was added to the medium at a concentration of 60 ng/mL. After 72 h of treatment, the medium was replaced with fresh media, and the cells were incubated for another 72 h. Total RNA was extracted using TRIzol® (ThermoFisher, Seoul, Korea) according to the manufacturer’s instructions, and cDNA was synthesized using PrimeScriptTM RT reagent Kit (Takara Bio, Shiga, Japan). qRT-PCR was conducted using TB Green® Premix Ex TaqTM (Tli RNaseH Plus, Takara Bio) with specific primers TRAP-F:5′-GCGACCATTGTTAGCCACATACGG-3′; TRAP-R:5′-CGCCCAGGGAGTCCTCAGATCCAT-3′ and primers for the housekeeping gene GAPDH-F:5′-AACTTT GGCATTGTGGAAGGGCTC-3′; GAPDH-R:5′-AAGGCCATGCCAGTGAGCTTC-3′. mRNA levels were normalized to those of GAPDH (Kim et al., 2015).
Four-week-old BALB/c mice were randomly assigned to the three groups (PBS, WT_CE, and mRANKL_CE). Before oral administration, 300 μL of neutralizing reagent (1.5% NaH2CO3) was administered to prevent gastric acidity. After 30 min, 100 μL 1×PBS was fed to the PBS group, and the cell extracts from 2.5×108 CFU wild type L. lactis IL1403 and cell extracts containing mRANKL (4.2 μg) from 2.5×108 CFU of recombinant L. lactis were fed to WT_CE and mRANKL_CE groups, respectively. All groups were fed for seven consecutive days and sampled on the eighth day (Supplementary Fig. S2).
Peyer’s patches of small intestine samples were extracted to determine the abundance of mature M cells by measuring the GP2 mRNA expression level. Isolation of RNA and synthesis of cDNA was same as above method. qRT-PCR was conducted using TB Green® Premix Ex TaqTM (Tli RNaseH Plus, Takara Bio) with M cell-specific primers GP2-F:5′- GATACTGCACAGACCCCTCCA-3′; GP2-R:5′- GCAGTTCCGGTCATTGAGGTA-3′ (Kusunose et al., 2020), and primers for the housekeeping gene GAPDH-F:5′-AACTTTGGCATTGTGGAAGGGCTC-3′; GAPDH-R:5′-AAGGCCATGCCAGTG AGCTTC-3′. mRNA levels were normalized to those of GAPDH (Kim et al., 2015).
Nine ileum samples of the same size (1 cm) were extracted from the same position (distal ileum). Total RNA was isolated from the tissues using the Maxwell (Promega) method. One milligram of total RNA was processed to prepare the mRNA sequencing library using the MGIEasy RNA Directional Library Prep Kit (MGI), and sequencing was performed using the MGIseq system.
After a quality check, raw reads were trimmed to 100 bp prior to mapping to the mouse reference genome GRCm38/mm10 using HiSat2 (version 2.1.0; Pertea et al., 2016). Following read alignment, counts assigned to features were computed using the featureCounts (version 2.0.3; Liao et al., 2014). Differentially expressed genes (DEGs) were identified using the DESeq2 (v1.24.0; Love et al., 2014). A gene was considered to be a DEG with a fold change>1.45 and adjusted p-value<0.1 using the Benjamini-Hochberg method (Benjamini and Hochberg, 1995). Gene Ontology (GO) enrichment analysis was performed using in-house Perl scripts. The significantly enriched GO terms were determined by Fisher’s exact test with p<0.05 and odds ratio>1.
Genomic DNA was extracted from fecal samples using a NucleoSpin Soil kit (Macherey-Nagel, Düren, Germany) according to the manufacturer’s instructions. DNA samples (5 ng) were used to amplify the 16S ribosomal RNA V4 region using Takara Ex-taq DNA polymerase (Takara Bio) with universal primer sets (Forward:5′-GGACTACHVGGGTWTCTAAT-3′ and R:5′-GTGCCAGCMGCCGCGGTAA-3′; Han et al., 2018). After amplification, all the samples were normalized to 50 ng per sample. A DNA library was then constructed and sequenced using the Illumina MiSeq platform (Illumina, San Diego, CA, USA), generating 2×300 bp paired-end reads.
To analyze the gut microbiome, de-multiplexed and pre-processed sequence reads were imported into Quantitative Insights Into Microbial Ecology (QIIME2, version 2021.2; Bolyen et al., 2019). Barcode and primer removal, quality control, amplicon sequence data correction, and de-replication, were performed using the DADA2 (Callahan et al., 2016). Feature tables and representative sequence files were merged for downstream analysis using QIIME2. Taxonomic classification was performed using the SILVA 132 database with 99% identity, based on the V4 16S region. All classification was performed within QIIME2 and was assigned using the naïve Bayesian algorithm available in the sklearn python library. For phylogenetic diversity analysis, alpha and beta diversities were calculated using the q2-diversity plugin, and included Faith’s phylogenetic diversity and weighted and unweighted UniFrac distances. Differential abundance analysis of microbiota was performed using an in-house Perl script. We considered a p-value<0.05 to indicate statistical significance.
Serum samples were collected on the eighth day, and concentration was measured using a Calcium Colorimetric Assay Kit (BioVision, Milpitas, CA, USA).
Statistical analysis was performed using an in-house Perl script and R (version 4.1.0) language. One-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test was used for determining statistical significance.
To confirm the expression of mRANKL, intracellular and secreted proteins were precipitated and analyzed by western blotting. No mRANKL was detected in the wild-type L. lactis samples. In recombinant L. lactis, intracellular and secreted mRANKL proteins were detected with expected size of 23.86 and 20.88 kDa, respectively (Fig. 1B). According to the standard curve, recombinant L. lactis produced 2.1 μg/mL of mRANKL in intracellular fraction (Supplementary Fig. S3).
To examine the physiological characteristics of wild type and the recombinant L. lactis, patterns of growth were measured. As shown in Supplementary Fig. S4, recombinant L. lactis showed a slightly delayed growth rate compared to the wild type; after 10 h though, these two strains showed similar growth patterns.
To validate the biological activity of mRANKL, RAW 264.7 cells were used to observe the stimulation of RANK-RANKL signaling. As shown in Supplementary Fig. S5, treatment with culture media containing commercial RANKL and mRANKL_CE significantly (p<0.05) increased TRAP mRNA expression levels compared with those in the PBS and WT_CE groups.
Cell extracts containing recombinant mRANKL were orally administered to mice for seven consecutive days, and the expression of mature M cells marker in Peyer’s patches was analyzed by qRT-PCR. As shown in Supplementary Fig. S6, the GP2 mRNA expression levels in the mRANKL_CE group were significantly (p<0.05) higher than those in the PBS and WT_CE groups. In addition, body weight gain was measured between day 1 and day 8 of the experiment, and the results showed that oral administration of WT_CE and mRANKL_CE had no significant effect on body weight (Supplementary Table S1). Our results showed that recombinant mRANKL from recombinant L. lactis is biologically active and does not cause significant changes in body weight.
To compare the gene expression between the PBS, WT_CE, and mRANKL_CE groups, RNA-Seq analysis of the mouse small intestine was performed. The WT_CE and mRANKL_CE groups were compared with the PBS group, with a cut-off fold change>1.45 and adjusted p-value<0.1. No DEGs were identified between the PBS and WT_CE groups. Between the PBS and mRANKL_CE groups, 63 DEGs, including 53 upregulated and 10 downregulated DEGs, were identified (Fig. 2 and Supplementary Table S2). Between the WT_CE and mRANKL_CE groups, 192 DEGs were identified, including 169 upregulated and 23 downregulated DEGs (Supplementary Table S2). These results indicate that cell extracts from WT_CE had no effect on mouse small intestinal gene expression, and only mRANKL had an effect.
To analyze the gene enrichment of DEGs, GO analysis were performed with a threshold of p<0.05 and odds ratio>1. Between the PBS and mRANKL_CE groups, 49 upregulated and 9 downregulated DEGs were annotated with GO terms. The upregulated GO terms included ‘de novo’ protein folding (p<0.001), response to unfolded protein (p<0.001), regulation of bone resorption (p<0.01), heat shock protein binding (p<0.001), and calcium ion binding (p<0.05; Table 1 and Supplementary Table S3). Between the WT_CE and mRANKL_CE groups, 159 upregulated and 22 downregulated DEGs were annotated with GO terms. The upregulated GO terms included calcium ion binding (p<0.001), calmodulin binding (p<0.001), and ‘de novo’ protein folding (p<0.05; Table 2 and Supplementary Table S3).
To compare the gut microbial diversity, the alpha and beta diversities of the three groups from normalized microbiome sequencing reads were investigated. For alpha diversity, three indices including observed features, Shannon and Faith’s phylogenetic diversity (Faith PD) were measured. None of the three indices showed any significant differences among the three groups (Supplementary Fig. S7). For beta diversity, principal coordinate analysis (PCoA) of unweighted and weighted UniFrac distances was performed. There were no significant differences among the three groups (Supplementary Fig. S8).
To compare the differences in major gut microbial taxa among the three groups, the microbial composition in these three groups were examined. The overall microbial composition in the gut was not significantly different among the three groups. However, the genera Lactobacillus (p<0.05), Sphingomonas (p<0.01) and Acinetobacter (p<0.01) differed significantly (Fig. 3 and Supplementary Table S4). Moreover, there was no significant difference among the three groups in the genus Lactococcus (Fig. 3 and Supplementary Table S4). These results showed that feeding cell extracts with or without mRANKL did not have a significant effect on the gut microbiome.
After RANKL administration, serum calcium levels were measured to compare the changes in calcium concentration among the three groups. There was no significant difference in serum calcium levels among the three groups (Supplementary Fig. S5).
Many recombinant microorganisms are widely used in food, chemical, and pharmaceutical industries today (Khan et al., 2016). The direct use of LMOs poses potential risks to the environment, humans, and animals (Prakash et al., 2011). LAB are traditionally used in fermentation and food preservation, and are recognized as safe for consumption. In particular, L. lactis has been engineered as a live vehicle for the delivery of DNA vaccines and production of therapeutic biomolecules (Tavares et al., 2020). In bacterial cell extracts study, cell extracts from recombinant L. lactis expressing SARS-CoV-2 spike protein were used to oral immunize mice, and antigen-specific antibodies were produced from immunized mice (Xuan et al., 2022). L. lactis IL1403 is a representative laboratory strain that can be used to produce recombinant heterologous proteins (Tavares et al., 2020), and there is no evidence to date that metabolites of L. lactis IL1403 are toxic to experimental animals. Moreover, use of cell-free extracts removes the risk of LMOs being released into the environment. Our results demonstrated that cell extracts containing mRANKL from recombinant L. lactis IL1403 can differentiate RAW 264.7 cells into osteoclast-like cells (Supplementary Fig. S5) and increase the number of mature M cells in mice (Supplementary Fig. S6). In addition, in the transcriptome analysis, no DEGs were found between PBS and WT_CE groups; while DEGs were found between PBS and mRANKL_CE groups. These results indicate that WT_CE has no effect on mouse small intestinal transcriptome, whereas mRANKL does have an effect. Moreover, RANKL not only affects the differentiation of M cells, but also affects other RANKL signal-responsive cells (Kukita and Kukita, 2013). Knoop et al. (2009) demonstrated not only GALT expressed RANK in mice (Knoop et al., 2009). Heat shock proteins are a family of proteins produced by cells in response to stress, and they positively regulate osteoclastic bone resorption through the RANKL-RANK signaling pathway (Hang et al., 2018). In this study, genes associated with heat shock protein binding and regulation of bone resorption were found to be upregulated in the mRANKL_CE group. Calcium signaling plays a significant role in osteoclastogenesis; the RANKL receptor utilizes calcium signaling to drive osteoclast differentiation (Komarova et al., 2003), and genes associated with calcium ion binding were found to be upregulated in the mRANKL_CE group. Additionally, SLIT3 was found to be co-upregulated in the mRANKL_CE group compared to its expression in the PBS (adjusted p<0.1) and WT_CE (adjusted p<0.1) groups (Supplementary Table S2). SLIT3 expression increases during osteoclast differentiation (Koh, 2018). In addition, bone remodeling-associated genes ZBTB16 (vs. PBS, adjusted p<0.001; vs. WT_CE, adjusted p<0.01) and ZBTB40 (vs. PBS, adjusted p<0.01; vs. WT_CE, adjusted p<0.1) were found to be co-upregulated in the mRANKL_CE group (Supplementary Table S2; Felthaus et al., 2014; Twine et al., 2016). However, no study so far has investigated whether an increase in RANKL levels in the intestine causes bone resorption. Since the cell extracts of the host producing recombinant proteins do not appear to have any effect on experimental animals, this strategy may be an alternative to the use of live cells. The well-known signal peptide USP45 is located at the upstream of the target protein and its secretion is relatively low. Improving the secretion yield of target proteins in cell-free culture supernatant, can be a good alternative to the direct use of LMOs. Recently, some studies have used postbiotics replace the live bacteria to improve the intestinal environment; postbiotics are defined as “non-viable bacterial products or metabolic products from microorganisms that have biological activity in the host” (Nataraj et al., 2020; Siciliano et al., 2021). Postbiotics contain a wide range of molecules, including peptidoglycans, surface proteins, cell wall polysaccharides, secreted proteins, bacteriocins, and organic acids, which can have positive effects on the host, including immunomodulatory, antitumor, antimicrobial, and barrier preservation effects (Nataraj et al., 2020; Teame et al., 2020). In our study, bacterial crude cell extracts had no adverse effect on the host, but exogenous proteins had favorable effects. Hence, exogenous protein can be produced by probiotics, and the cell extract can then be used for therapeutic or other purposes. Also, the crude cell extracts can be used directly, without the need for purification (Taghinezhad-S et al., 2021). There have been many recent studies on the host-microbiome interactions in postbiotics (Nataraj et al., 2020; Peluzio et al., 2021; Siciliano et al., 2021; Teame et al., 2020). However, there have been no studies on the state of intestinal microbiome after intake of live recombinant LAB or its cell extracts. In this study, NGS were used to characterize how cell extracts of wild-type or recombinant L. lactis affected the gut microbiome. Although there was no significant change in the gut microbiome, the abundances of several genera were significantly different (Fig. 3 and Supplementary Table S4). RANKL stimulates the differentiation of monocyte / macrophage precursor cells into osteoclasts, and overexpression of RANKL resulting in bone erosion in rheumatoid arthritis (RA; Tanaka, 2019). The abundance of Lactobacillus is significantly higher in patients with RA than in healthy controls (Li et al., 2021). Our results also showed that the abundance of Lactobacillus was significantly (p<0.05) higher in the mRANKL_CE group. The abundance of Sphingomonas was also significantly (p<0.01) higher in the mRANKL_CE group. Eriksson et al. (2022) showed that Sphingomonas abundance was positively correlated with RA (Eriksson et al., 2022). Another RA-related genus, Bacteroides, showed decreased abundance in the mRANKL_CE group, although the difference was not significant. In addition, the abundance of Bacteroides was decreased in patients with RA compared with its abundance in the healthy controls (Wang et al., 2022). These results may indicate changes in abundances of specific microbes caused by excessive RANKL in the intestine. One diagnostic marker in RA patients is a higher serum calcium concentration than that in the healthy controls (Tawfik et al., 2019). In addition, an increase in RANKL levels in the bone microenvironment leads to bone resorption and increased calcium release (Ono et al., 2020). Our result indicated no significant differences in serum calcium levels among the three groups (Supplementary Table S5). Moreover, the abundance of Lactococcus, which was used as the host for the production of mRANKL, was not significantly different among the three groups (Fig. 3). This indicates that the cell extracts of L. lactis did not elicit an immune response, which may be because Lactococcus is a natural inhabitant of the mouse gut. Our results indicate that the cell extracts of L. lactis did not have a significant impact on the mouse gut microbiome.
In summary, the cell extracts containing mRANKL from recombinant L. lactis are biologically active both in vitro and in vivo, and the cell extracts from WT_CE did not affect intestinal transcriptome. In addition, there was no significant change in the gut microbiome after administration of cell extracts of WT_CE or recombinant L. lactis. This strategy could potentially be used for the development of protein drugs.
Supplementary materials are only available online from: https://doi.org/10.5851/kosfa.2022.e54. |
|
PMC9647186 | Xing Wang,Cheng Chung Yong,Sejong Oh | Metabolites of Latilactobacillus curvatus BYB3 and Indole Activate Aryl Hydrocarbon Receptor to Attenuate Lipopolysaccharide-Induced Intestinal Barrier Dysfunction | 01-11-2022 | Latilactobacillus curvatus BYB3,aryl hydrocarbon receptor,Caco-2 cells,tight junctions,lipopolysaccharide | Abstract This study aimed to investigate the effects of the metabolites of Latilactobacillus curvatus BYB3 and indole-activated aryl hydrocarbon receptor (AhR) to increase the tight junction (TJ) proteins in an in vitro model of intestinal inflammation. In a Western blot assay, the metabolites of L. curvatus BYB3 reduced the TJ demage in lipoploysaccharide (LPS) stimulated-Caco-2 cells. This reduction was a result of upregulating the expression of TJ-associated proteins and suppressing the nuclear factor-κB signaling. Immunofluorescence images consistently revealed that LPS disrupted and reduced the expression of TJ proteins, while the metabolites of L. curvatus BYB3 and indole reversed these alterations. The protective effects of L. curvatus BYB3 were observed on the intestinal barrier function when measuring transepithelial electrical resistance. Using high-performance liquid chromatography analysis the metabolites, the indole-3-latic acid and indole-3-acetamide concentrations were found to be 1.73±0.27 mg/L and 0.51±0.39 mg/L, respectively. These findings indicate that the metabolites of L. curvatus BYB3 have increasing mRNA expressions of cytochrome P450 1A1 (CYP1A1) and AhR, and may thus be applicable for therapy of various inflammatory gut diseases as postbiotics. | Metabolites of Latilactobacillus curvatus BYB3 and Indole Activate Aryl Hydrocarbon Receptor to Attenuate Lipopolysaccharide-Induced Intestinal Barrier Dysfunction
This study aimed to investigate the effects of the metabolites of Latilactobacillus curvatus BYB3 and indole-activated aryl hydrocarbon receptor (AhR) to increase the tight junction (TJ) proteins in an in vitro model of intestinal inflammation. In a Western blot assay, the metabolites of L. curvatus BYB3 reduced the TJ demage in lipoploysaccharide (LPS) stimulated-Caco-2 cells. This reduction was a result of upregulating the expression of TJ-associated proteins and suppressing the nuclear factor-κB signaling. Immunofluorescence images consistently revealed that LPS disrupted and reduced the expression of TJ proteins, while the metabolites of L. curvatus BYB3 and indole reversed these alterations. The protective effects of L. curvatus BYB3 were observed on the intestinal barrier function when measuring transepithelial electrical resistance. Using high-performance liquid chromatography analysis the metabolites, the indole-3-latic acid and indole-3-acetamide concentrations were found to be 1.73±0.27 mg/L and 0.51±0.39 mg/L, respectively. These findings indicate that the metabolites of L. curvatus BYB3 have increasing mRNA expressions of cytochrome P450 1A1 (CYP1A1) and AhR, and may thus be applicable for therapy of various inflammatory gut diseases as postbiotics.
Intestinal epithelial cells (IECs) with intact tight junctions (TJs) form a barrier between the external environment and the mammalian host (Yu et al., 2018). Normal functioning of the intestinal epithelial barrier is critical for maintaining health (Citi, 2018; Odenwald and Turner, 2017; Turner, 2009). Disruption of TJs and paracellular permeability can promote the entry of molecules and activate the immune system, leading to continuous tissue destruction (Lee, 2015). Hence, maintaining the integrity of the intestinal epithelial barrier is critical for inhibiting the development of gastrointestinal diseases and inflammation (Tlaskalová-Hogenová et al., 2004). Indole, an interspecies and interkingdom signaling molecule, plays essential roles in bacterial pathogenesis and eukaryotic immunity (Lee et al., 2015). The human intestinal tract is rich in a diverse range of about 1014 commensal bacteria, some of which are crucial for nutrient assimilation and benefit the immune system (Tlaskalová-Hogenová et al., 2004). A metabolomic study demonstrated that the production of indoxyl sulfate and the antioxidant indole-3-propionic acid in animal blood depended entirely on enteric bacteria (Wikoff et al., 2009). In addition, indole and its derivatives may influence human diseases, such as bacterial infections, intestinal inflammation, neurological diseases, diabetes, and cancers (Lee et al., 2015). Multiple protein complexes, which are crucial components of TJs, are located in IECs (Tsukita et al., 2001) and include occludin, claudins, and zonula occludens (ZO). These protein complexes are vital for the maintenance of TJs and permit cytoskeletal regulation of the intestinal barrier integrity (Van Itallie and Anderson, 2006). Pathogens damage the intestinal epithelial barrier, increase intestinal permeability, and induce the development of inflammatory bowel disease (IBD) and necrotizing enterocolitis (NEC; Guo et al., 2015). IBD includes two chronic idiopathic inflammatory diseases, ulcerative colitis, and Crohn’s disease (Arrieta et al., 2009). Lipopolysaccharide (LPS) is a harmful antigen that can trigger inflammatory responses in the intestinal tissue and can be detected in the serum of patients with NEC and IBD (Han et al., 2020). Recent studies have identified the association between clinically relevant concentrations (1–10 ng/mL) of LPS and intestinal barrier dysfunction under in vivo and in vitro conditions (Guo et al., 2013). In our previous study, Latilactobacillus curvatus BYB3 decreased the disease activity score of dextran sulfate sodium-induced colitis in a mouse model (Wang et al., 2022). Supplementation with indole or using Lactobacillus reuteri with high aryl hydrocarbon receptor (AhR) ligand production can improve some metabolic symptoms (Swimm et al., 2018). Therefore, we hypothesized that L. curvatus BYB3 has a similar function. The supernatants of L. curvatus BYB3 and the metabolites of L. curvatus BYB3+indole ameliorated LPS-induced intestinal barrier dysfunction by upregulating the levels of TJ proteins in Caco-2 cells. These findings illustrated the mechanism underlying the destructive effect of clinically relevant concentrations of LPS on the intestinal epithelial barrier, providing evidence for the clinical application of metabolites of L. curvatus BYB3+indole in the treatment of LPS-induced intestinal barrier dysfunction. The AhR is a ligand-dependent transcription factor that is widely expressed in vertebrates and is involved in numerous biological processes, such as cell proliferation (Xie et al., 2012), apoptosis (Marlowe et al., 2008), differentiation (Xie et al., 2012), and inflammatory response (Neavin et al., 2018). The AhR separates from its molecular chaperone complex and forms a heterodimer with the aryl hydrocarbon nuclear translocator (ARNT) in the nucleus. This AhR-ARNT dimer then binds to the upstream regulatory region of its target genes, such as the cytochrome P450 family 1 genes (CYP1A1 and CYP1B1; Esser and Rannug, 2015). Indoles may have utility as an intervention to limit the decline of barrier integrity and the resulting systemic inflammation that occurs with aging (Powell et al., 2020). Indoles and indole-metabolites secreted by the commensal bacteria have been shown to extend the healthspan of diverse organisms, including Caenorhabditis elegans, Drosophila melanogaster, and mice. The effects of indole and metabolites on animal heatlthspan were found to be AhR-mediated (Sonowal et al., 2017). This study was conducted to research the effect of L. curvatus BYB3 on the intestinal epithelial barrier of the Caco-2 cells. Furthermore, we investigated the differences in mRNA expression levels of CYP1A1 and AhR in response to the metabolites of L. curvatus BYB3 and indole.
LPS derived from Escherichia coli O111:B4 was purchased from Sigma-Aldrich (Burlington, MA, USA) and dissolved in phosphate-buffered saline (PBS) to prepare the stock solutions with concentrations of 1 mg/mL. DL-indole-3-lactic acid (ILA), 3-indoleacetic acid (IAA), indole-3-acetamide (IAM), and indole were purchased from Sigma-Aldrich, and trifluoroacetic acid was procured from Daejung Chemicals and Metals (Siheung, Korea). All reagents were stored as specified by the manufacturer. The following antibodies were used in the study: Zona occludens 1 (ZO-1) antibody (Cat No.21772-1-AP, Proteintech, Chicago, IL, USA), claudin-1 antibody (Cat No. ab211737, Abcam, Cambridge, UK), nuclear factor-kappa B (NF-κB) antibody (sc-372, Santa Cruz Biotechnology, Dallas, TX, USA), p-NF-κB p65 (Cell Signaling Technology, Danvers, MA, USA), β-actin C4 antibody (sc-4778, Santa Cruz Biotechnology, Dallas, TX, USA), secondary R-antibody (Cat. No. A11036, Invitrogen, Waltham, MA, USA), and Westar Supernova (Code.XLS3,0100, Cyanagen, Srl, Bologna, Italy).
Twenty-one probiotic candidates (Table 1) were cultivated in MRS medium (DifcoTM Lactobacilli MRS broth, BD Diagnostics, Franklin Lakes, NJ, USA) for 24 h at 37°C. Before use, the overnight LABs were diluted to a cell density of 107 CFU/mL in MRS broth prior to use. Indole was added at a final concentration of 58.5 mg/mL, and the cells were incubated at 37°C for 24 h. Then samples were centrifuged at 2,719.5×g for 15 min at room temperature. A total of 1 mL of the supernatant was collected and mixed immediately with 0.4 mL of Kovac’s reagent to determine the extracellular indole concentration. After the Kovac’s reagent was added, the mixture was vortexed to separate the phases. The top phase was collected, and the absorbance was measured at 540 nm.
L. curvatus BYB3 cells were isolated from traditional homemade kimchi in Gwangju and Jeollanam-do, and maintained in MRS broth. Cells were incubated in the MRS broth at 37°C and centrifuged at 1,500×g for 15 min at room temperature to obtain cell pellets. The pellets were stored in 10% glycerol or skim milk at –80°C until further use. The supernatant was filtered using a 0.2 μm syringe (Sartorius AG, Gottingen, Germany). The cells in the indole group were treated with 58.5 mg/mL indole, those in the BYB3 group were incubated with L. curvatus BYB3, and those in the BYB3+indole group were treated with both L. curvatus BYB3 and 58.5 mg/mL indole. The three groups were incubated in the MRS medium for 24 h at 37°C. The cell pellets were discarded, and the CFSs were used to treat the Caco-2 cells.
The Caco-2 cells used in the study were obtained from the Korean Cell Line Bank (No.30037.1, Seoul, Korea). Caco-2 cells were cultured in Modified Eagles Medium (MEM), high glucose (HyClone Laboratories, Logan, UT, USA), supplemented with 20% fetal bovine serum (GibcoTM, Thermo Fisher Scientific, Waltham, MA, USA), 1% MEM non-essential amino acids solution (100×; GibcoTM, Thermo Fisher Scientific), and 1% antibiotic-antimycotic solution (GibcoTM, Thermo Fisher Scientific) at 37°C in an atmosphere containing 5% CO2. The medium was replaced every two or three days. The Caco-2 cells (1×106 cells) were seeded in a 20×90 mm dish and treated with 10 ng/mL of LPS. They were then treated with 1 mL of indole, 1 mL of L. curvatus BYB3, and 1 mL of the BYB3+ indole metabolites supernatants.
Caco-2 cells (1×103 cells/cm2) were seeded in a Corning®, Costar®, Transwell® chamber with 0.4 μm pores (Corning, New York, NY, USA) that had been placed in a 24-well plate. Another Transwell® plate was kept blank. After reaching confluence, the cells were differentiated and polarized for 7–10 days in the culture medium. Subsequently, the Caco-2 cells were treated with 10 ng/mL of LPS and later with 100 μL of indole supernatant, 100 μL of L. curvatus BYB3 supernatant, and 100 μL of BYB3+ indole supernatants. The TEER assay was used to measure cell monolayer integrity before and after all treatments. The TEER was measured using an epithelial volt-ohm-meter equipped with a chopstick electrode [Millicell® ERS-2 (Electrical Resistance System), EMD Millipore, Burlington, MA, USA]. The electrode was immersed at a 90° angle, with one tip in the basolateral chamber and the other in the apical chamber. Care was taken to prevent contact of the electrode with the monolayer. Measurements were performed in triplicate for each monolayer. An insert without Caco-2 cells was used as a blank; the mean resistance of the blank was subtracted from all samples. The unit area resistance was calculated by dividing the resistance values by the effective membrane area (0.33 cm2).
Caco-2 cells (1×106 cells) seeded in a 20×90 mm dish were treated with 10 ng/mL of LPS followed by treatment with 1 mL of indole supernatant, 1 mL of strain 3, 15 and LGG supernatant, and 1 mL of indole+strain 3, 15 and LGG supernatants. After incubation for 24 h, the cells were collected for further analysis. Total RNA was isolated and converted into complementary DNA (cDNA) as described previously. Briefly, 2 μg of total RNA was used to cDNA using a Maxime RT PreMix kit (Oligo Dt primer; Cat. No. 25081, iNtRON Biotechnology, Seongnam, Korea). The following primers were used for real-time polymerase chain reaction (RT-PCR; Table 2). PCR was performed under the following conditions: Initial denaturation at 94°C for 3 min, followed by 40 cycles of the program with incubations at 94°C for 30 s, 60°C for 30 s, and 72°C for 1 min, followed by incubation at 65°C for 5 s, until the end of the program. The relative gene expression levels were determined by comparative analyses using the formula:
Caco-2 cells (1×106 cells) were seeded in a 20×90 mm dish and treated with 10 ng/mL of LPS and then 1 mL of the indole supernatant, 1 mL of BYB3 supernatant, and 1 mL of indole+BYB3 metabolites supernatant. After incubation for 24 h, the cells were collected for further analysis. The total protein concentration in the cell lysates was determined using the PRO-PREP protein extraction solution (iNtRON Biotechnology). Briefly, 5×106 cells were immersed in 400 μL of the PRO-PREP solution and homogenized in ice for 10–20 min. The mixture was then centrifuged at 13,000×g at 4°C for 5 min, and the extracted protein was collected in the supernatant. The protein concentration was determined by the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Equal amounts of protein (50 μg per lane) were separated using 10% sodium dodecyl sulfate-polyacrylamide gel, electroblotted (Mini-PROTEAN® II Cell Systems, Bio-Rad Laboratories, Hercules, CA, USA), and transferred to a polyvinylidene difluoride membrane (Bio-Rad Laboratories). The proteins were blocked with 5% skim milk (Difco Laboratories, Detroit, MI, USA) and underwent overnight antibody incubation against E-cadherin, N-cadherin, Vimentin, and β-actin at 4°C. After incubation, the membranes were washed and incubated with horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit antibodies for 1 h at room temperature. After each was washed three times with PBST for 10 min, protein bands developed. The bands were detected via enhanced chemiluminescence, and the band density was determined using β-actin as the reference protein.
Caco-2 cells were seeded on a 24-well plate at a density of 1×103 cells/mL. These cells were treated with 10 ng/mL of LPS and then with 100 μL each of indole, L. curvatus BYB3, BYB3+indole metabolite supernatants. After treatment for 24 h, the cells were collected for the next step. The cells were prepared as described in Material and Methods. Caco-2 cells were grown on glass coverslips; the slides were washed with PBS for 5 min at room temperature, fixed with 3.7% formaldehyde in PBS buffer for 20 min at 4°C, and again rinsed thrice with PBS buffer for 5 min at room temperature. The monolayers were permeabilized with 0.5% TritonTM X-100 (Sigma-Aldrich) for 20 min at room temperature and rinsed three times with PBS buffer for 2 min at room temperature. The slides were blocked with 5% skim milk in tris buffered saline with Tween® (TBST) for 1 h at room temperature without rinsing. They were then incubated with rabbit polyclonal anti-ZO-1 antibody, rabbit monoclonal anti-claudin-1 antibody, and rabbit polyclonal anti-NF-κB p65 antibody for 2 h at room temperature. The slides were rinsed thrice with TBST for 5 min at room temperature. The remaining incubations were performed in the dark. The slides were further incubated with an Alexa Fluor® 568 goat anti-rabbit secondary antibody (Abcam). Nuclei were stained using 4’,6-diamidino-2-phenylindole dihydrochloride (DAPI; Cat. No D1306, Invitrogen) for 15 s at room temperature. The samples were covered with a coverslip using the Vectashield® anti-fade mounting medium (Vector cat. #H-1000, Vector Laboratories, Newark, CA, USA). The edges of the coverslips were sealed by nail polishing. The slides were examined and analyzed using a fluorescence microscope (Olympus BX50, Olympus, Tokyo, Japan).
The indole derivatives in the CFSs were analyzed as previously described. Briefly, filtered samples were injected (10 mL), in triplicate, into an HPLC system (KNAUER, Wissenschaftliche Geräte GmbH, Berlin, Germany) equipped with a C-18 gravity 150×4.6 mm column, particle size: 5 μm (Macherey-Nagel GmbH & Co. KG, Düren, Germany). The flow rate was set to 1 mL/min, and the column oven temperature was maintained at 30°C. The running buffers were 0.3% trifluoroacetic acid solutions prepared in ultra-pure water (A) and acetonitrile (B). The process was initiated with an A:B ratio of 90:10; the linear gradient was applied to reach this ratio in 1 min. The steps included gradients with 55% solution A: 45% solution B for 28 min, 5% solution A: 95% solution B for 30 and 35 min, and 90% solution A: 10% solution B for 36 min. The measurements were stopped after 45 min. The detection wavelength was set at 280 nm.
All data are presented as mean±SD of triplicate experiments. Statistical significance comparing different sets of groups was determined using the Student’s t-test. In experiments comparing multiple experimental groups, statistical differences between groups were analyzed using one-way analysis of variance (ANOVA). Statistical analyses were performed using IBM® SPSS® Statistics 20 (IBM, Chicago, IL, USA), and a p<0.05 was considered statistically significant.
The probiotic candidates (Table 1 and Fig. 1) show the Lactobacillus strains’ ability to metabolize and reduce indole concentration during fermentation. Among the tested Lactobacillus strains, 3 (L. curvatus BYB3), 15 (Lactobacillus acidophilus), and 21 (Lactilacobacillus rhamnosus GG) demonstrated remarkable indole reducing abilities and were selected for subsequent analyses.
Caco-2 cells were treated with the 10% MRS as the control and 10% supernatant of the strains (3, 15, and 21) previously screened for 24 h. Treatment of the Caco-2 cells with 10 ng/ mL of LPS simulated the conditions of colitis. A previous study detected increased expression of CYP1A1, which is indicative of the AhR activation (Yu et al., 2018). To confirm the activation of the AhR by the metabolites, the mRNA expression levels of CYP1A1 and AhR were determined after treating the Caco-2 cells for 24 h with LPS alone or in combination with other supernatants. The supernatants of L. curvatus BYB3 and indole significantly increased the mRNA expression CYP1A1 and AhR by 35-fold and 3-fold, respectively (Figs. 2A and B).
The TEER was used to measure cell monolayer integrity, which was assessed before and after all treatments. LPS increased the permeability of the intestinal epithelial barrier. However, the effects of the metabolites of L. curvatus BYB3 and indole on the LPS-mediated increase in intestinal permeability are unknown. LPS significantly decreased the TEER after 12 h; the reduction continued for 24 h after application (Fig. 3). In contrast, the metabolites of L. curvatus BYB3 and indole remarkably increased the TEER. This finding suggests that the metabolites reduced the permeability of the intestinal epithelial barrier. In addition, the supernatants of L. curvatus BYB3 and indole increased the TEER. However, co-treatment with the metabolites and LPS significantly restored the LPS-mediated increase in the permeability of the intestinal epithelial barrier in Caco-2 cells (Figs. 2 and 3). Hence, these metabolites could significantly protect against LPS-induced intestinal permeability.
LPS down-regulated the expression of the ZO-1, occludin, and claudin-1 proteins. Caco-2 cells were co-treated with 10 ng/mL of LPS, the supernatants of indole and L. curvatus BYB3, and the metabolites of BYB3+indole for 24 h to determine alterations in the expression of the TJ proteins. The Caco-2 cells treated with L. curvatus BYB3+indole showed increased expression of the ZO-1 and claudin-1 proteins compared to cells treated with LPS alone (Figs. 4A and B). Furthermore, cells treated with the metabolites of BYB3+indole showed a significant increase in the expression of ZO-1 and claudin-1. To explore the anti-inflammatory effects of the metabolites on Caco-2 cells, alterations in NF-κB, a biomarker of inflammation, were examined. NF-κB p65 and the protein levels of total and phospho-p65 were detected by Western blot analysis. Compared to the LPS-treated cells (10 ng/mL, control), the Caco-2 cells treated with the supernatants of L. curvatus BYB3, indole, and metabolites of BYB3+indole showed decreased NF-κB expression. The reduction was significant in the presence of the metabolites of L. curvatus BYB3+indole. Interestingly, co-treatment with the metabolites of L. curvatus BYB3+indole and LPS had a remarkable effect on the attenuation of LPS-induced inflammation.
Immunofluorescence was used to detect the localization and expression of TJ proteins, as these results were more intuitive. The LPS-treated group showed severe disruption in the structure of TJ proteins structure (Fig. 5). In contrast, the TJ protein ZO-1 was intact without any damage in the cells treated with LPS+BYB3+indole. LPS-induced disruption was repaired in the LPS+indole, and LPS+BYB3 treated groups. Examination of claudin-1 expression revealed a trend similar to that observed for ZO-1 (Figs. 5A and B). Consistent with this observation, immunofluorescence analysis of NF-κB demonstrated that p65 accumulated within the nucleus of Caco-2 cell monolayers treated with the metabolites of L. curvatus BYB3+indole. However, incubation with LPS decreased LPS-induced nuclear accumulation of NF-κB (Fig. 5C).
HPLC analysis was performed to precisely identify and quantify indole derivatives. Several indole derivatives (100 μM each) were separated, and their peaks were detected by HPLC using a C-18 reverse column (Fig. 6). Under optimal conditions, the retention times of IAM, ILA, IAA, and indole were 13.2, 16.1, 19.3, and 29.1 min, respectively (Fig. 6A). The peak in Fig. 6B corresponds to the main components because of the presence of indole from the supernatants of the indole-treated 0 h. The three peaks in Fig. 6C represent IAM, ILA, and indole. The main component in Fig. 6C, indicated by three peaks, including two of IAM and one of ILA, represented the supernatant of the L. curvatus BYB3 group fermented for 24 h. The indole content in the supernatants of the L. curvatus BYB3+indole group was reduced, and IAM and ILA metabolites were observed to varying degrees (Fig. 6D).
Indole alleviates the symptoms of gastrointestinal disorders by activating the AhR (Hubbard et al., 2015). Several compounds have been proposed as putative endogenous AhR ligands, many of which are produced via pathways involved in the metabolism of tryptophan and indole, including indole-3-aldehyde, IAA, and many more (Bittinger et al., 2003; Chung and Gadupudi, 2011). In our previous study, AhR activation inhibited NF-κB expression, in vivo and in vitro (Salisbury and Sulentic, 2015). In macrophages, the activation of AhR signaling blocks NF-κB binding sites and masks NF-κB transcription activity, suppressing NLRP3 inflammasome activation (Huai et al., 2014). Hence, the current study aimed to identify the potential effect of metabolites of L. curvatus BYB3 and indole in mediating the recovery of TJ after LPS-induced disruption of the intestinal barrier in the colon mucosal cell layer. Our preliminary studies showed that L. curvatus BYB3 might play a role in alleviating inflammatory responses. However, the association between intestinal TJ proteins and inflammation influenced by L. curvatus BYB3 was not elucidated under in vitro conditions. The findings from this study suggest that the metabolites of L. curvatus BYB3 and indole can activate the AhR. In previous studies, LPS-induced inflammation disrupted the integrity of IECs and increased paracellular permeability (Gao et al., 2017). The results from this study demonstrated that the supernatants of L. curvatus BYB3, indole, and metabolites of BYB3+indole inhibited LPS-induced inflammation in IECs by enhancing the expression of TJ proteins and decreasing paracellular permeability in Caco-2 cells. However, direct evidence is required to explore the association between the supernatants of cells treated with L. curvatus BYB3, indole, and the metabolites of L. curvatus BYB3+indole and intestinal permeability; such evidence was not available earlier. ZO-1, occludin, and claudin-1 are important TJ proteins that maintain permeability in the small intestine (Anderson and Van Itallie, 1995). Western blot analysis revealed that the administration of the metabolites of L. curvatus BYB3+indole significantly improved intestinal epithelial barrier function by increasing the expression of the TJ proteins ZO-1 and claudin-1. Deregulated NF-κB activation has been previously reported to contribute to the pathogenesis of various inflammatory diseases (Liu et al., 2017). In this study, the metabolites of BYB3+indole decreased NF-κB expression. In a previous study, we determined that Lactobacillus improved the intestinal epithelial barrier function by increasing the expression of TJ proteins (Zeng et al., 2020). TEER is a commonly used indicator of intestinal epithelial membrane permeability (Srinivasan et al., 2015). An increase in the TEER and a decrease in paracellular permeability reflect the enhancement of the barrier function (Capaldo et al., 2017). The small intestine is one of the main organs of the digestive system, and the Caco-2 cell monolayer is a recognized intestinal cell line. According to this study’s HPLC analysis, only three indole compounds were detected among the metabolites of L. curvatus BYB3+indole, namely IAM, indole, and ILA. Several Bacteroides spp. and Clostridium bartlettii have been reported to produce ILA and IAA, whereas Bifidobacterium spp. have been reported to produce ILA (Aragozzini et al., 1979; Russell et al., 2013). However, there are few reports of L. curvatus producing ILA. Our results provide evidence that microbiota-mediated metabolism inhibits LPS-induced inflammation, increasing the expression of TJ proteins. Based on the primary research results of this study, key metabolic molecules that improve intestinal should be investigated in further studies. We demonstrated that the metabolites of L. curvatus BYB3 and indole inhibited LPS-induced inflammation in IECs by enhancing TJs, which, in turn, reduced paracellular permeability. HPLC results confirmed that various concentrations of indole and indole derivatives (ILA and IAM) enhance TJ protein expression. This protective effect may provide a potential approach to restoring TJ barrier function, and AhR may be a novel therapeutic target in gut health and diseases such as IBD. |
|
PMC9647188 | Won-Young Bae,Woo-Hyun Jung,So Lim Shin,Seulgi Kwon,Minn Sohn,Tae-Rahk Kim | Investigation of Immunostimulatory Effects of Heat-Treated Lactiplantibacillus plantarum LM1004 and Its Underlying Molecular Mechanism | 01-11-2022 | Lactiplantibacillus plantarum,postbiotics,immunostimulatory effect,nuclear factor kappa B | Abstract Postbiotics are defined as probiotics inactivated by heat, ultraviolet radiation, sonication, and other physical or chemical stresses. Postbiotics are more stable than probiotics, and these properties are advantageous for food additives and pharmacological agents. This study investigated the immunostimulatory effects of heat-treated Lactiplantibacillus plantarum LM1004 (HT-LM1004). Cellular fatty acid composition of L. plantarum LM1004 isolated form kimchi was analyzed by gas chromatography–mass spectrometry detection system. The nitric oxide (NO) content was estimated using Griess reagent. Immunostimulatory cytokines were evaluated using enzyme-linked immunosorbent assay. Relative protein expressions were evaluated by western blotting. Phagocytosis was measured using enzyme-labelled Escherichia coli particles. L. plantarum LM1004 showed 7 kinds of cellular fatty acids including palmitic acid (C16:0). The HT-LM1004 induced release of NO and upregulated the inducible NO synthase in RAW 264.7 macrophage cells. Tumor necrosis factor-α and interleukin-6 levels were also increased compared to control (non-treated macrophages). Furthermore, HT-LM1004 modulated mitogen-activated protein kinase (MAPK) subfamilies including p38 MAPK, extracellular signal-regulated kinase 1/2, and c-Jun N-terminal kinase. Therefore, these immunostimulatory effects were attributed to the production of transcriptional factors, such as nuclear factor kappa B (NF-κB) and the activator protein 1 family (AP-1). However, HT-LM1004 did not showed significant phagocytosis of RAW 264.7 macrophage cells. Overall, HT-LM1004 stimulated MAPK/AP-1 and NF-κB expression, resulting in the release of NO and cytokines. These results will contribute to the development of diverse types of food and pharmacological products for immunostimulatory agents with postbiotics. | Investigation of Immunostimulatory Effects of Heat-Treated Lactiplantibacillus plantarum LM1004 and Its Underlying Molecular Mechanism
Postbiotics are defined as probiotics inactivated by heat, ultraviolet radiation, sonication, and other physical or chemical stresses. Postbiotics are more stable than probiotics, and these properties are advantageous for food additives and pharmacological agents. This study investigated the immunostimulatory effects of heat-treated Lactiplantibacillus plantarum LM1004 (HT-LM1004). Cellular fatty acid composition of L. plantarum LM1004 isolated form kimchi was analyzed by gas chromatography–mass spectrometry detection system. The nitric oxide (NO) content was estimated using Griess reagent. Immunostimulatory cytokines were evaluated using enzyme-linked immunosorbent assay. Relative protein expressions were evaluated by western blotting. Phagocytosis was measured using enzyme-labelled Escherichia coli particles. L. plantarum LM1004 showed 7 kinds of cellular fatty acids including palmitic acid (C16:0). The HT-LM1004 induced release of NO and upregulated the inducible NO synthase in RAW 264.7 macrophage cells. Tumor necrosis factor-α and interleukin-6 levels were also increased compared to control (non-treated macrophages). Furthermore, HT-LM1004 modulated mitogen-activated protein kinase (MAPK) subfamilies including p38 MAPK, extracellular signal-regulated kinase 1/2, and c-Jun N-terminal kinase. Therefore, these immunostimulatory effects were attributed to the production of transcriptional factors, such as nuclear factor kappa B (NF-κB) and the activator protein 1 family (AP-1). However, HT-LM1004 did not showed significant phagocytosis of RAW 264.7 macrophage cells. Overall, HT-LM1004 stimulated MAPK/AP-1 and NF-κB expression, resulting in the release of NO and cytokines. These results will contribute to the development of diverse types of food and pharmacological products for immunostimulatory agents with postbiotics.
Lactic acid bacteria (LAB), regarded as useful probiotics, play a crucial role in fortifying the intestinal barrier against food-borne pathogenic bacteria (Kao et al., 2020; Levit et al., 2019) and modulating intestinal microbiota and immune systems (Levit et al., 2019). LAB have been consumed in various types of foods, including dairy products (Oshiro et al., 2021; Parvarei et al., 2021), fermented fruits and vegetables (Lorn et al., 2021; Oshiro et al., 2021), sourdough (Oshiro et al., 2021), and meat products (Parlindungan et al., 2021). Currently, LAB are used as pharmaceutical agents and not limited to probiotics (Barros et al., 2020). Postbiotics, which are inactivated probiotics (Barros et al., 2020; Parvarei et al., 2021) and their metabolites, have been investigated in a broad spectrum of food and pharmaceutical industries (Barros et al., 2020). Innate immune system operates as the first-line defense in the host (Lee et al., 2020). Myeloid cells (macrophages, monocytes, and neutrophils) are critical components of innate immunity (Mantovani and Netea, 2020). Myeloid cells recognize pathogen-associated molecular patterns (PAMPs) from infectious microbes and danger-associated molecular patterns from injured tissues caused by Toll-like receptors (TLRs), retinoic acid-inducible gene I-like receptors, nucleotide-binding oligomerization domain-like receptor family proteins, and absent in melanoma 2, a family of pattern-recognition receptors (PRRs; Lee et al., 2020). Innate immune system activates macrophages and immediately counteracts to pathogens to provide host defenses against various types of invaders (Geng et al., 2018; Jeong et al., 2019; Lee et al., 2020; Leopold Wager and Wormley, 2014; Liu et al., 2019; Mantovani and Netea, 2020; Netea et al., 2020; Um et al., 2020). Unlike these rapid reactions, the trained immune system involves reprogramming of innate immune cells awakened by exogenous or endogenous stimulations (Netea et al., 2020). For example, monocytes that were first treated with β-glucan lost stimulus within 24 h, and the second challenge by lipopolysaccharides (LPS) showed a burst of tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6; Bauer et al., 2018). These well-trained cells undergo epigenetic modifications and have long-term memory effects (Bauer et al., 2018; Netea et al., 2020). Lactiplantibacillus plantarum is a facultative heterofermentative Lactobacillus species (Liu et al., 2018). The European Food Safety Authority (EFSA) has acknowledged L. plantarum as a Qualified Presumption of Safety (QPS) and has continuously updated its strains since 2007 (Andreoletti et al., 2008; Liu et al., 2018). L. plantarum has been used not only for starter culture in the food industry (Le and Yang, 2018; Liu et al., 2018) but also in pharmacological research owing to its bio-functionalities (Le and Yang, 2018). The aim of this study is investigation of immunostimulatory effects of heat-treated L. plantarum LM1004 (HT-LM1004). Other studies have been focused on the immunostimulatory effects of probiotics and their metabolites, such as exopolysaccharides, whereas heat-treated probiotics are not of interest. It had been known that heat treatment disrupts the bacterial cell wall and induces release of nucleic acid, peptidoglycan, and teichoic acids, resulting in modulating immune responses by these strain specific bacterial components (Piqué et al., 2019). Our previous study, micronized and heat-treated L. plantarum (MHT-LM1004) and HT-LM1004 showed increase of natural killer (NK) cell activity and relative cytokine production in immune-suppressed mice (Jung et al., 2019). However, molecular level mechanisms of immunostimulatory effect of HT-LM1004 were not fully understood. In addition, MHT-LM1004 is defined as similar material due to manufacturing methods and is not suitable for industrial production. Therefore, the immunostimulatory effects of HT-LM1004 via the mitogen-activated protein kinase (MAPK)/Activator protein 1 family (AP-1)/Nuclear factor kappa B (NF-κB) pathway were investigated in this study.
LPS from E. coli O111:B4 (LPS) and ammonium pyrrolidine dithiocarbamate (APDC) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Thiazolyl Blue tetrazolium bromide (MTT) was obtained from Alfa Aesar (Haverhill, MA, USA). Antibodies against cyclooxygenase-2 (COX-2), phospho-p38 MAPK, p38 MAPK, phospho-extracellular signal-regulated kinase 1/2 (ERK1/2), ERK1/2, phospho-c-Jun N-terminal kinase (JNK), JNK, c-Fos, c-Jun, phospho-IκBα, IκBα, phospho-p65 NF-κB, p65 NF-κB, phospho-AMP-activated protein kinase (AMPK), AMPK, phospho-acetyl-CoA carboxylase (ACC), ACC, and glyceraldehyde-3-phosphate dehydrogenase were obtained from Cell Signaling Technology (Danvers, MA, USA). The anti-inducible nitric oxide synthase (iNOS) antibody was obtained from GeneTex (Irvine, CA, USA).
L. plantarum LM1004 was isolated from kimchi, Korean traditional fermented food. In brief, 25 g of kimchi was homogenized in 225 mL of phosphate buffered saline (PBS) using a stomacher. After homogenization, sample was diluted in peptone water (0.1%, w/v) and spread on de Man-Rogosa-Sharpe (MRS) agar (for Lactobacillus; BD, Franklin Lakes, NJ, USA), M17 agar (for lactic Streptococcus and Lactococcus; MBcell, Seoul, Korea), and Bifidobacterium selective agar (Bifidobacterium spp.; MBcell). The spread agar plates were incubated at 37°C for 48 h. After 48 h, colonies isolated from MRS agar were spread on Bromocresol purple (BCP) agar and yellow colonies on BCP agar further purified in newly prepared MRS agar until single colony. Single and pure colony was enriched in MRS broth for gram-staining and catalase reaction. The isolate was identified gram-positive and catalase-negative strain with rod-type shape. Isolated strain was named LM1004 and identified by 16S rRNA sequencing as L. plantarum. L. plantarum LM1004 was stored in MRS containing with 20% glycerol at –80°C until use (Ngamsomchat et al., 2022). For analysis of complete genome sequencing, genomic DNA (gDNA) of L. plantarum LM1004 was extracted by TaKaRa MiniBEST Bacteria Genomic DNA Extraction Kit (Takara Bio, Kusatsu, Japan). The DNA sequencing library was constructed using single molecular real-time sequencing technology (Pacific Biosciences, Menlo Park, CA, USA). De novo assembly was performed using Celera Assembler in hierarchical genome assembly process (Macrogen, Seoul, Korea).
HT-LM1004 was obtained from the Department of Production, Lactomason (Jinju, Korea). The cell numbers and morphology were constantly managed by the Quality Management Team (Lactomason). Lyophilized heat-treated cells were assigned the product number 11NTF8 and stored at –20°C until use.
Extraction of cellular fatty acid from HT-LM1004 was performed by Bligh and Dyer method with modification (Cheng et al., 2022). Briefly, 200 μL of chloroform/methanol solution (2:1, v/v) and 300 μL of 0.6 M hydrochloric acid solution (in methanol) were added in 20 mg of HT-LM1004. The mixture was shaken for 2 min vigorously and heated at 85°C during 60 min. The extracted lipids were cooled at 25°C for 20 min. Fatty acid methyl esters (FAME) were more extracted by n-hexane for 60 to 120 min. The FAME extracted layer (n-hexane layer) was transferred into clear vial and stored at –20°C until analysis. Cellular fatty acid analysis was performed by gas chromatography–mass spectrometry detection (GC/MSD). The GC/MSD system was composed of Agilent 8890 gas chromatography system coupled with a 5977B mass selective detector and 7693A automated liquid sampler (Agilent, Santa Clara, CA, USA). An Agilent J&W DB-FastFAME capillary column packed with cyanopropyl (30 m×0.25 mm, 0.25 μm) was employed. Injection port temperature was 250°C in constant flow and 1 μL of sample was injected using the spilt mode of 20:1. Ultrapure helium gas was used as carrier gas with a flow rate of 1 mL/min. The initial oven temperature was retained at 60°C for 1 min, raised from 60°C to 165°C at a rate of 60°C/min, held 1 min at 165°C, raised form 165°C to 230°C at a rate of 5°C/min, and kept for 3 min. The temperature of ion source and transfer line was 230°C, and 250°C, respectively. The mass spectra were obtained on an electron ionization at 70 eV and recorded m/z 40–550 of mass range. Methyl undecanoate was used as internal standard (Liu et al., 2022).
The murine macrophage cells, RAW 264.7 cell lines, were obtained from the Korean Cell Line Bank (KCLB, Seoul, Korea). The cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin solution at 37°C in a humidified atmosphere containing 5% CO2. When the cells were grown to 80% confluence, they were gently harvested using a scraper. Harvested cells were seeded in various well plates and incubated for 24 h. After incubation, cells were treated with LPS (10 ng/mL) or HT-LM1004 to measure immunostimulatory and phagocytic effects (Liu et al., 2019). Cells were pre-treated with APDC, an NF-κB inhibitor, for 1 h before treatment with LPS or HT-LM1004.
Macrophage cells were seeded in a 96-well microplate (1×105 cells/well) and incubated for 24 h. After incubation, each well was treated with LPS or HT-LM1004 (1×107, 2.5×107, 5×107, and 1×108 cells/mL) and further incubated for 24 h. The incubated cells were washed with PBS twice times and fresh DMEM including 0.5 mg/mL of MTT was used to replace any lost medium (Um et al., 2020). The absorbance of MTT formazan which viable cell converted was evaluated at 570 nm using a microplate reader (SpectraMax iD3, Molecular Devices, San Jose, CA, USA). Cell viability was calculated as follows: where Asample is the absorbance of LPS or HT-LM1004 treated cells and Acontrol is the absorbance of non-treated cells (negative control).
The NO content was measured using Griess reagent (Geng et al., 2018). Briefly, RAW 264.7 macrophage cells (1×105 cells/well) were treated with LPS (positive control) or HT-LM1004 for 24 h. After 24 h, cell-free supernatants were collected, and added Griess reagent (Promega, Madison, WI, USA) for measuring NO contents according to the manufacturer’s guidelines.
The release of immunostimulatory cytokines (TNF-α and IL-6) was measured using an enzyme-linked immunosorbent assay (Bo et al., 2019; Liu et al., 2019). Cell culture and sample treatments were prepared as described in cell culture and treatment. Cell-free supernatants were collected by centrifugation at 1,000×g for 20 min at 4°C. All immunostimulatory cytokines were analyzed according to the manufacturer’s guidelines (Invitrogen, Waltham, MA, USA).
The phagocytic effect of HT-LM1004 treated RAW 264.7 macrophage cells was evaluated using enzyme-labeled E. coli particles (CytoSelectTM 96-Well Phagocytosis Assay, Cell Biolabs, San Diego, CA, USA; Jeong et al., 2019). Relative phagocytic effects were measured by enzyme-substrate reactions, according to the manufacturer’s guidelines.
The HT-LM1004 treated RAW 264.7 macrophage cells were lysed by RIPA lysis buffer (containing 50 mM Tris-HCl, 150 mM NaCl, 1% Triton-X, 1% sodium deoxycholate, 0.1% SDS and 2 mM EDTA) with protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific, Waltham, MA, USA). The lysed cells were centrifuged at 13,000×g for 20 min at 4°C. The supernatants were collected, and protein content was measured using the PierceTM BCA Protein Assay Kit (Thermo Fisher Scientific). The extracted proteins were stored at 4°C until further use. Proteins were separated by capillary western blot analysis (Khan et al., 2021) or sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). For capillary western blotting, proteins were diluted to 0.8 mg/mL and separated by a 12–230 kDa capillary cartridge (ProteinSimple, San Jose, CA, USA) according to the manufacturer’s guidelines. Protein separation and immunodetection were conducted using JESS, an automated western blotting system (ProteinSimple). For SDS-PAGE, proteins (30 μg) were separated to 8% or 10% of SDS-PAGE gel and transferred onto polyvinylidene fluoride membranes. Membranes were blocked with 3.75% skim milk for 1 h, followed by incubation with primary antibodies at 4°C for 24 to 48 h. After incubation, the membranes were washed with Tris-buffered saline containing Tween 20 and incubated with secondary antibodies at 25°C for 2 h. The blots were visualized using ECL detection reagent (Advansta, San Jose, CA, USA). The intensity of the bands was analyzed using ImageJ software.
Statistical analyses were performed using SPSS Statistics version 18 software (IBM, Armonk, NY, USA). HT-LM1004 treated groups were compared with the negative control (non-treated group). The mean values were analyzed using a t-test at p<0.05.
The whole genome characteristics of L. plantarum LM1004 are shown in Fig. 1A. The size of entire gene sequence of L. plantarum LM 1004 was 3,198,690 bp, single and circular chromosome with 44.59% of GC content. A total of 3001 protein-coding sequences were identified in L. plantarum LM1004. The chromosome include 16 rRNA and 68 tRNA. The complete genome sequence of L. plantarum LM1004 has been deposited in the National Center for Biotechnology Information GenBank database under the accession number CP025988. Cellular fatty acid composition of HT-LM1004 are shown in Figs. 1B and C. Total 7 kinds of fatty acid were observed in HT-LM1004. Lactobacillic acid (cycC19:0) and palmitic acid (C16:0) were investigated most abundant cellular fatty acid in HT-LM1004. The proportion of saturated fatty acid (SFA), unsaturated fatty acid (USFA) and cyclic fatty acid (CFA) in HT-LM1004 were measured 41.42%, 20.03%, and 38.55%, respectively.
Prior to measuring the immunostimulatory effects of HT-LM1004, cell viability was investigated using MTT formazan. Cytotoxicity was not shown in HT-LM1004 or LPS treated (positive control) macrophage cells (Fig. 2A). The NO content and iNOS expression are shown in Figs. 2B and C. The HT-LM1004 (1×107, 2.5×107, 5×107, and 1×108 cells/mL) treated RAW 264.7 macrophage cells released 3.05, 7.55, 12.55, and 16.32 μM of NO, respectively (p<0.01). The relative expression of iNOS increased 6.59-, 14.24-, 17.14-, and 19.86-fold compared to non-treated RAW 264.7 macrophages (negative control; p<0.01).
The release of immunostimulatory cytokines (TNF-α and IL-6) and relative protein expression of COX-2 are shown in Fig. 3. HT-LM1004 increased TNF-α secretion from 205.52 (negative control) to 1,530.11, 1,925.27, 3,445.44, and 3,906.01 pg/mL (p<0.01). In addition, HT-LM1004 treated RAW 264.7 macrophages released 254.36, 302.66, 394.29, and 651.93 pg/mL of the immunostimulatory cytokine IL-6 (p<0.01). The relative protein ex-pression of COX-2 was up-regulated in HT-LM1004 treated RAW 264.7 macrophage cells. COX-2 expression increased to 25.25-fold at 1×108 cells/mL of HT-LM1004 treated RAW 264.7 macrophage cells (p<0.001; Fig. 3C).
Figs. 4 and 5 present changes in MAPK and transcription factor in HT-LM1004 treated RAW 264.7 macrophage cells. HT-LM1004 treated RAW 264.7 macrophage cells were used to investigate the phosphorylation of MAPK sub-families (p38, ERK1/2, and JNK). Briefly, phosphorylation of p38 MAPK, ERK1/2, and JNK increased to 4.96-, 5.52-, and 2.98-fold at 1×108 cells/mL of HT-LM1004 treated macrophage cells (p<0.05). Moreover, phosphorylation of IκBα and activation of NF-κB p65 translocation were observed in HT-LM1004 treated cells (Fig. 5; p<0.01). Other transcription factors (c-Fos and c-Jun) also increased protein expression (p<0.05).
APDC, a pharmacological NF-κB inhibitor, prevents iNOS expression and NO pro-duction (Dong et al., 2015). In the current study, the immunostimulatory effects of HT-LM1004 were investigated by the upregulation of iNOS and the release of NO in APDC-treated RAW 264.7 macrophage cells. The APDC-treated cells inhibited the release of NO (3.81 μM) though HT-LM1004 treated cells produced 4.46 and 7.31 μM of NO at 5×107 and 1×108 cells/mL, respectively (p<0.01). The APDC and HT-LM1004 co-treated cells also showed an over-expressed iNOS level comparing to non-treated RAW 264.7 macrophage cells (p<0.001; Fig. 6).
The phagocytic effect of HT-LM1004 treated cells is shown in Fig. 7. The 1×107 cells/mL of HT-LM1004 treatment increased phagocytosis of macrophage cells (123.18%), but no significant differences were detected. Phosphorylation of AMPK and ACC did not significantly change in the HT-LM1004 treated macrophages.
The interactions between LAB and the host immune system have not been clearly reported, but many researchers have suggested that PRRs recognize LAB cell wall-derived molecules as PAMPs (Ren et al., 2020). Lipoteichoic acid (LTA), the most representative cell wall-derived PAMP in gram-positive bacteria, is an important ligand for innate immune responses (Friedrich et al., 2022; Jung et al., 2022; Kang et al., 2011; Ren et al., 2020). LTA is an amphiphilic molecule with both a hydrophilic polysaccharide moiety and a hydrophobic glycolipid region (Kang et al., 2011). In general, LTA interacts with TLR2, which is associated with myeloid differentiation primary response 88 (MyD88), interleukin-1 receptor-associated kinases, and TNF receptor-associated factor 6 (Jung et al., 2022). These TLR2-MyD88 dependent signaling pathways upregulate release of immunostimulatory cytokines and chemokines (Kang et al., 2011). The immunogenicity of LTA depends on its structural diversity in accordance with the genus and species levels (Friedrich et al., 2022). Ryu et al. (2009) reported that LTA isolated from three different gram-positive bacteria (Staphylococcus aureus, Bacillus subtilis, L. plantarum) showed relative differences in NF-κB translocation and TNF-α secretion. Moreover, Jung et al. (2022) reported differential immunostimulatory effects of LTA isolated from four different strains of L. plantarum and analyzed differences in glycolipid composition. Considering the immunogenicity of LTA, heat-treated LAB also showed immunomodulatory effects of LTA. In the current study, HT-LM1004 induced the release of immunostimulatory cytokines (TNF-α and IL-6; Fig. 3) and translocation of NF-κB (Fig. 5). These results were also observed in other heat-treated L. plantarum species (Choi et al., 2018; Jeong et al., 2019; Moon et al., 2019). In addition, Kim et al. (2018) reported that heat-treated LAB contributed immunomodulatory food additives and prolonged the shelf life. NO is synthesized in various cells for neurotransmission, vascular function, host defense, and immune regulation (Xue et al., 2018). NO synthases are classified into three subtypes: Neuronal NO synthase, endothelial NO synthase, and iNOS. In particular, iNOS is mainly expressed in immune-stimulated cells by cytokines and inflammatory molecules (PAMPs) such as LPS and LTA (Kang et al., 2011; Xue et al., 2018). NO plays a critical role in the regulation of M1 macrophage polarization. M1 macrophages are able to respond to pro-inflammatory responses and produce cytokines such as TNF-α, IL-6, and IL-12 for host defense (Yunna et al., 2020). Additionally, NO generated by iNOS expression in M1 macrophages directly defends against pathogens (Xue et al., 2018). HT-LM1004 induced release of NO levels and expressed iNOS in RAW 264.7 macrophage cells (Fig. 2). HT-LM1004 also affected the release of immunostimulatory cytokines (Fig. 3). NF-κB signaling is crucial in physiological processes. NF-κB transcription factors are involved in cellular transformation and proliferation, apoptosis, angiogenesis, metastasis, and activation of the immune system (Aggarwal, 2004). In the immune system, the NF-κB transcription factor is involved in inflammatory responses to microbes and viruses by innate immune cells and the development of adaptive immune cells in secondary lymphoid organs (Dorrington and Fraser, 2019). IκBα degradation and phosphorylation activates the NF-κB transcription factor (p65) from the cytoplasm to the nucleus (Geng et al., 2018). The translocation of NF-κB mediates the transcription of immunostimulatory molecules and cytokines, including iNOS, COX-2, TNF-α, IL-2, IL-6, and IL-12 (Geng et al., 2018; Moon et al., 2019; Yang et al., 2019). TNF-α, IL-2, and IL-12 contribute to the activation of NK cells, which play a crucial role in the host defense system against pathogens and transformed cells (Lauwerys et al., 2000; Moon et al., 2019). These cytokines promote cytotoxicity of NK cells and immunomodulatory effects of NK cells in the innate and adaptive immune systems (Lauwerys et al., 2000). In addition, Sharma and Das (2018) reported that IL-2 mediates the proliferation of NK cells. The MAPK cascade promotes the transcription of transcriptional factors, such as NF-κB and AP-1 (Geng et al., 2018; Liu et al., 2019; Yang et al., 2019). AP-1 consists of four subfamilies: Jun, Fos, ATF-activating transcription factor protein families and musculoaponeurotic fibrosarcoma. AP-1 in immune system has been reported to play a role in Th differentiation, T-cell activation, and T-cell anergy (Atsaves et al., 2019). Activation of the MAPK pathway via a cascade of phosphorylation events on serine/threonine residues coordinates downstream of AP-1 and NF-κB (Atsaves et al., 2019; Geng et al., 2018). Thus, MAPK activation by HT-LM1004 plays a central role in the innate immune system. Phagocytosis is occurred in three classes of phagocytic cells in immune system such as monocytes/macrophages, neutrophilic granulocytes and dendritic cells (Schumann, 2016). Macrophages act as scavenger of pathogens, dead cells, and debris. When macrophages engulf pathogens, phagosomes are fused with lysosomes which result in phagolysosomes and toxic peroxides for digesting the pathogens (Jeong et al., 2014; Schumann, 2016). Fatty acids can influence modulating of immune response of macrophages including phagocytosis and cytokine productions (Schumann, 2016). Calder et al. (1990) reported SFA such palmitic acid result in decrease of phagocytosis of macrophages. On the other hand, USFA increased phagocytosis of macrophages except oleic acid (C18:1). However, palmitic acid activate TLR-MyD88 dependent NF-κB activation and production of immunostimulatory cytokines (Korbecki and Bajdak-Rusinek, 2019). In current study, the contents of SFA were measured 2-times higher than USFA in cellular fatty acid of HT-LM1004. Palmitic acid is a most abundant fatty acid except for lactobacillic acid which is CFA (Figs. 1B and C). The 1×107 cell/mL of HT-LM1004 treated cells showed highest phagocytosis effect (123.18%) while 1×108 cell/mL treated macrophage cells decreased to 116.69% (Fig. 7). However, palmitic acid derived from cellular membrane of HT-LM1004 induced immunostimulatory effects by activation of NF-κB (Figs. 5 and 6).
In the present study, the immunostimulatory potency of HT-LM1004 was investigated at various stages of innate immunity. HT-LM1004 stimulated the MAPK pathway and regulated transcription factors, such as AP-1 (c-Fos and c-Jun) and NF-κB p65. These transcription factors induce secretion of NO, TNF-α, and IL-6 to enhance the immune system. Heat-treated LAB lost their probiotic properties, but HT-LM1004 showed immunostimulatory effects as a postbiotic. These results suggest HT-LM1004 as an immunostimulatory agent, food additive, and therapeutic agent. |
|
PMC9647189 | Mengyi Zhao,Ruiming Yue,Xiaoxiao Wu,Zhan Gao,Miao He,Lingai Pan | The diagnostic value of metagenomic next-generation sequencing for identifying Pneumocystis jirovecii infection in non-HIV immunocompromised patients | 27-10-2022 | metagenomic next-generation sequencing (mNGS),Pneumocystis jirovecii (P. jirovecii),infection,transplantation,immunocompromised | Background Pneumocystis jirovecii pneumonia (PJP) remains an important cause of morbidity and mortality in non-HIV immunocompromised patients especially in transplant recipients. But its diagnosis remains challenging due to the insuffificient performance of conventional methods for diagnosing Pneumocystis jirovecii(P. jirovecii) infection. Therefore, the auxiliary diagnostic function of metagenomics next-generation sequencing (mNGS) in clinical practice is worth of exploring. Method 34 non-HIV immunocompromised patients who were diagnosed as PJP by clinical manifestations, imaging findings, immune status of the host, and Methenamine silver staining were tested by mNGS from October 2018 to December 2020 in Sichuan Provincial People’s Hospital. The clinical performances of mNGS for P. jirovecii infection diagnosis were also evaluated with genome reads abundance and comparing with other traditional diagnostic methods. Results We diagnosed a total of 34 non-HIV PJP patients by the clinical composite diagnosis. Our data shows that, compared with the clinical microbiological test, the detection rate of mNGS for P. jirovecii in non-HIV infected PJP patients is significantly higher than that of Methenamine silver staining and serum 1-3-β-D-glucan. mNGS can be used as an auxiliary diagnostic tool to help diagnosis. The number of reads mapped to the genome of P. jirovecii and the duration of patients from onset to sampling collection were statistically significant between the two groups (Reads>100 and Reads ≤ 100) (8days vs. 23days, p=0.020). In addition, univariate analysis showed that C-reactive protein (15.8mg/L vs.79.56mg/L, p=0.016), lactate dehydrogenase (696U/l vs. 494U/l, p=0.030) and procalcitonin (0.09ng/ml vs. 0.59ng/ml, p=0.028) was also statistically significant between the two groups. Conclusions An effective detection rate was achieved in PJP patients using mNGS testing of bronchoalveolar lavage fluid (BALF) or blood. The study also confirmed that the abundance of reads of P. jirovecii is related to the interval between the onset and sample collection. And the inflammation status during simultaneous mNGS detection might determine the abundance of pathogens. Hence, we conclude that the mNGS strategy could benefit disease diagnosis as well as treatment when complicated clinical infections appeared. | The diagnostic value of metagenomic next-generation sequencing for identifying Pneumocystis jirovecii infection in non-HIV immunocompromised patients
Pneumocystis jirovecii pneumonia (PJP) remains an important cause of morbidity and mortality in non-HIV immunocompromised patients especially in transplant recipients. But its diagnosis remains challenging due to the insuffificient performance of conventional methods for diagnosing Pneumocystis jirovecii(P. jirovecii) infection. Therefore, the auxiliary diagnostic function of metagenomics next-generation sequencing (mNGS) in clinical practice is worth of exploring.
34 non-HIV immunocompromised patients who were diagnosed as PJP by clinical manifestations, imaging findings, immune status of the host, and Methenamine silver staining were tested by mNGS from October 2018 to December 2020 in Sichuan Provincial People’s Hospital. The clinical performances of mNGS for P. jirovecii infection diagnosis were also evaluated with genome reads abundance and comparing with other traditional diagnostic methods.
We diagnosed a total of 34 non-HIV PJP patients by the clinical composite diagnosis. Our data shows that, compared with the clinical microbiological test, the detection rate of mNGS for P. jirovecii in non-HIV infected PJP patients is significantly higher than that of Methenamine silver staining and serum 1-3-β-D-glucan. mNGS can be used as an auxiliary diagnostic tool to help diagnosis. The number of reads mapped to the genome of P. jirovecii and the duration of patients from onset to sampling collection were statistically significant between the two groups (Reads>100 and Reads ≤ 100) (8days vs. 23days, p=0.020). In addition, univariate analysis showed that C-reactive protein (15.8mg/L vs.79.56mg/L, p=0.016), lactate dehydrogenase (696U/l vs. 494U/l, p=0.030) and procalcitonin (0.09ng/ml vs. 0.59ng/ml, p=0.028) was also statistically significant between the two groups.
An effective detection rate was achieved in PJP patients using mNGS testing of bronchoalveolar lavage fluid (BALF) or blood. The study also confirmed that the abundance of reads of P. jirovecii is related to the interval between the onset and sample collection. And the inflammation status during simultaneous mNGS detection might determine the abundance of pathogens. Hence, we conclude that the mNGS strategy could benefit disease diagnosis as well as treatment when complicated clinical infections appeared.
Pneumocystis jirovecii pneumonia (PJP) is a common opportunistic infection in the immunocompromised population (Eddens and Kolls, 2015). PJP cases have been reported frequently in solid organ transplant recipients, particularly in renal transplant recipients (Yiannakis and Boswell, 2016; Chen et al., 2020; Le Gal et al., 2020), and other patients like cancer, patients with congenital or acquired immunodeficiency or patients treated with immunosuppressive drugs, and so on (Wickramasekaran et al., 2017; Chen et al., 2020). With the progress of medical technology, the population of organ transplantation or immunosuppression is gradually increasing, so the population of PJP may also increase. Compared with the human immunodeficiency virus (HIV)-PJP, the early clinical symptoms of non-HIV PJP are atypical and nonspecific, and more likely to lead to alveolar damage and respiratory failure, and rapidly develop into severe pneumonia, resulting in extremely high mortality. (Monnet et al., 2008; Tasaka et al., 2010; Roux et al., 2014). At present, the incidence of PJP in the non-HIV population continues to increase, which deserves attention (Cillóniz et al., 2019). Rapid pathogen diagnosis and accurate treatment play a key role in improving outcomes in non-HIV patients with PJP (Zhang et al., 2021). Since P. jirovecii cannot grow stably in vitro, the detection tools for P.jirovecii, such as microscopy and Gomori’s methenamine silver staining, have limitations with low positive rate. Because these traditional methods require a high pathogen burden in the lungs and experienced microbiologists to ensure microscopic detection rates of P. jirovecii, which may lead to a certain false negative (Ma et al., 2018; Le Gal et al., 2020; Jiang et al., 2021). Besides, real-time PCR based diagnostic tests have significantly improved sensitivity and specificity, while commercial kits have not yet been widely used in clinical in China (Summah et al., 2013; Zhang et al., 2021). Metagenomic next-generation sequencing (mNGS) can complement traditional diagnostic methods through high-throughput sequencing, which can directly detect nucleic acids from pathogens in clinical samples, and then analyze nucleic acid sequences through bioinformatics methods. Comparing with traditional methods, mNGS has a higher detection rate of PJP (Wang et al., 2019; Zhang Y et al., 2019; Irinyi et al., 2020; Li et al., 2020). In addition, compared with other traditional diagnostic methods, the turnaround time of mNGS would be accomplished within 48 hours, therefore, mNGS can play an advantage in clinically auxiliary diagnosis (Somasekar et al., 2017; Chiu and Miller, 2019). mNGS could simultaneously identify bacteria, fungi, viruses, and various parasites from clinical samples such as cerebrospinal fluid, plasma, tissue, pleural fluid, etc. in an unbiased, synchronized, and straightforward manner (Yao et al., 2016; Chen et al., 2021). In recent years, the feasibility of mNGS in the pathogen identifications from clinical samples has been confirmed (Miao et al., 2018).There are some kinds of pathogens, such as fungi or viruses are not detectable by conventional culture methods, but can be detected by mNGS accurately and showed a higher positive detection rate (Charpentier et al., 2017; De La Cruz and Silveira, 2017). Therefore, nowadays, the implementation of mNGS becomes a very important detection method for severe or emerging pathogen infection. To validate our main hypothesis that mNGS would perform better in emerging infection agents identification, in this study, the clinical data, routine biochemical tests, microbial culture, and mNGS results of 34 non-HIV immunocompromised PJP patients admitted to Sichuan Provincial People’s Hospital were analyzed to evaluate the application value of mNGS in clinical diagnosis.
We included patients who were highly suspicious of PJP from 2018 to 2020 in Sichuan Provincial People’s Hospital. At present, the reference standard for diagnosis of PJP is still mainly to find characteristic cysts and trophozoites through staining and microscopic examination of clinical specimens, combined with the immune status of the host, degree of immunosuppression, imaging characteristics, and other biochemical indicators are taken as the auxiliary diagnosis of PJP. In this study, we combined the above mentioned factors as the reference standard for diagnosing PJP. We didn’t conduct PCR, because this detection method is not commonly used in most clinical microbiology laboratories in China (Bandt and Monecke, 2007). Patients were eligible for enrollment if they met all the following criteria:1) Immunocompromised, such as hematological malignancies, solid organ transplantation, hematopoietic stem cell transplantation, rheumatic immune system diseases, long-term use of corticosteroids or immunosuppressants, skin system diseases, etc.; 2) Typical clinical manifestations include subacute attacks, progressive dyspnea, accompanied by symptoms such as fever, dry cough, dyspnea and fatigue, and progressive hypoxemia; 3) Highresolution chest CT showed typical diffuse reticular nodules in both lungs, beginning with hilar nodular interstitial infiltration, mainly diffuse ground-glass opacities, with occasional plaques and consolidations; ([Chinese guidelines for diagnosis and treatment of HIV/AIDS (2018)], 2018; Fishman and Gans, 2019; Wang et al., 2019). 34 immunocompromised patients were diagnosed as PJP based on the host clinical status and immune states, clinical features, imaging findings, the results of Methenamine silver staining, the result of mNGS, and the comprehensive judgment of two senior clinicians in the hospital. The flow diagram of the study is described in Supplemental Figure S1 . Clinical data of all confirmed cases were recorded, including demographic characteristics, sample collection, and clinical microbiological test results. After admission, bronchoalveolar lavage fluid (BALF) or blood was sent for mNGS test.
The volume of 3-4 mL of whole blood was drawn from patients, placed in the blood collection tube(BD Biosciences, State of New Jersey, USA) and stored at room temperature for 3-5 minutes before plasma separation and centrifuged at 4,000 rpm for 10 min at 4°C(Eppendorf, Hamburg, Germany)within 8 h of collection. Plasma samples were transferred to new sterile tubes(Gene Era Biotech, California, USA).DNA was extracted from 300 uL of plasma using the TIANamp Micro DNA Kit (DP316, TIANGEN BIOTECH, Beijing, China) following the manufacturer’s operational manual. The extracted DNA specimens were used for the construction of DNA libraries (Long et al., 2016).
1.5-3mL BALF sample from the patient was collected according to standard procedures. 1.5mL microcentrifuge tube(Gene Era Biotech, California, USA) with 0.6mL sample and 1g 0.5mm glass bead were attached to a horizontal platform on a vortex mixer(Thermo, Massachusetts, USA) and agitated vigorously at 2800-3200 rpm for 30 min. 0.3mL sample was separated into a new 1.5mL microcentrifuge tube and DNA was extracted using the TIANamp Micro DNA Kit (DP316, TIANGEN BIOTECH) according to the manufacturer’s recommendation.
Then, the DNA library was constructed by DNA fragmentation, end repair of blunt ends caused by fragmentation, ligation of adapters for amplification and sequencing, and PCR amplification to enrich the target fragments using the MGIEasy Cell-free DNA Library Prep Set (MGI Tech, Shenzhen, China), according to the instructions of the manual. Agilent 2100 was used for quality control of the DNA libraries. Quality qualified libraries were sequenced by the BGISEQ-50/MGISEQ-2000 platform (Jeon et al., 2014).
High-quality sequencing data were generated by removing low-quality reads, followed by computational subtraction of human host sequences mapped to the human reference genome (Hg19) using Burrows-Wheeler Alignment (Li and Durbin, 2009). The remaining data by removal of low-complexity reads were classified by simultaneously aligning to four Microbial Genome Databases, consisting of bacteria, fungi, viruses, and parasites. The classification reference databases were downloaded from NCBI (ftp://ftp.ncbi.nlm.nih.gov/genomes/). Reference sequences contained 4,945 whole genome sequences of virus, 6,350 bacterial genomes or scaffolds, 1064 fungi related to human infection, and 234 parasites associated with human diseases. The Genbank accession number of P. jirovecii reference genome sequences using for the mapping is GCA_001477535.1.
BALF and blood were smeared on glass slides. After drying naturally, the samples were stained with silver staining solution (Zhuhai beso Biotechnology Co., Ltd., Shenzhen, China) according to the operation instructions. After waiting 3-5 min, P. jirovecii was microscopically examined for cyst structure. The cell wall components of fungi(1-3-β-D-glucan) were detected according to the instructions of the fungal (1-3)-β-D-glucan detection (chromogenic method) kit (Dana Biotechnology Co., Ltd., Tianjin, China).
All continuous variables were expressed as medians, the Mann-Whitney U test was used to compare the differences of continuous variables between the two groups, continuous variables with a P value 0.05 were considered statistically significant, and all tests were two-tailed. All statistical analyses were performed using SPSS 22.0 software.
In this study, P. jirovecii was identified in the BALF/blood of 34 patients by mNGS, who were diagnosed as PJP. Of these confirmed cases, 22 were male (64.7%) and 12 were female (35.3%). The mean age was 51.79 years (from 20 years to 84 years), 19 were kidney recipients, 1 liver recipient, 5 cases of connective tissue disease, 4 cases of blood system diseases, and 5 cases of others (including tumors and skin diseases). The detection results of sample types, the number reads of P. jirovecii, 1-3-β-D-glucan, and Methenamine silver staining are shown in Table 1 . The patient had no significant adverse events following imaging studies, mNGS, and other diagnostic procedures.
Among 34 patients who were diagnosed with PJP by mNGS, the number of reads mapped to P. jirovecii genome ranged from10 to 239032 and the mean number of reads for P. jirovecii was 2534. BALF/blood samples for mNGS were collected from 1 to 159 days after suspected patient infection, with a median collection time of 16.5 days, and it was found that there was no statistical difference. Based on the number of reads of P. jirovecii, we divided 34 patients into 2 groups, one group with reads ≤ 100(7 patients, 20.59%)and the other with reads>100(27 patients, 79.41%). Comparing the two groups, the duration from onset to sampling collection was generally shorter in patients with reads>100 than in those with reads ≤100 (8 days vs. 23 days, p=0.020). There was a statistically significant difference in the symptom of cough (7 persons vs. 14 persons, p=0.021),the symptom of sputum(5 persons vs. 7 persons, p=0.027), and clinical indicators such as C-reactive protein (15.8mg/L vs. 79.56mg/L, p=0.016), lactate dehydrogenase(696U/l, 494U/l, p=0.030) and procalcitonin (0.09ng/ml vs. 0.59ng/ml, p= 0.028) ( Table 2 ).
Methenamine silver staining is the most commonly used clinical test for PJP, and serum 1-3-β-D-glucan is a widely used serological marker for PJP, if the value of 1-3-β-D-glucan is elevated, the patient may be infected with P. jirovecii. In the present study, we compared the diagnostic value of the mNGS, Methenamine silver staining and serum 1-3-β-D-glucan in 34 PJP patients. P. jirovecii was detected from patients by clinical microbiological testing methods, of which 34 patients were detected by Methenamine silver staining, only 1 was positive (2.9%) and 34 patients were detected by 1-3-β-D-glucan, and the value greater than 60 pg/mL was 67.65% (23/34). However, we detected P. jirovecii in 34 patients (100%) by mNGS. Thus the study shows that mNGS has a high detection rate for P. jirovecii infections, which also confirmed that the clinical mNGS performance is helpful for infection diagnosis.
Currently, the incidence of PJP in non-HIV immunocompromised populations is increasing due to the prevalence of immunosuppressive diseases such as hematological malignancies, solid tumors, systemic corticosteroid therapy, immunosuppressive therapy, and organ transplantation (Gaborit et al., 2019). It has also been shown that in non-HIV-infected people, PJP usually develops into respiratory failure within a short period and results in a 30–60% mortality rate, which is significantly higher than in HIV-infected people (Cillóniz et al., 2019). Therefore, the early diagnosis of P. jirovecii infections in non-HIV patients is particularly important. In the past, the diagnosis of P. jirovecii was mainly based on staining, PCR, and detection of 1-3-β-D-glucan is a common cell wall constituent of most pathogenic fungi, including P. jirovecii (Lu et al., 2011a). But these methods have their limitations. Routine staining requires a large number of pathogens in the lungs and an experienced microbiologist to ensure the detection of P. jirovecii under the microscope; therefore, it can be insensitive and biased (Procop et al., 2004). Studies have demonstrated that respiratory specimen PCR results are sufficient to confirm or rule out disease in high-risk patients with suspected P. jirovecii (Lu et al., 2011b), but it might fail value in single-shot detection of mixed infections, especially for those rare or novel strains (White et al., 2017). In addition, PCR and microarrays use specific primers or probes to target only one or a limited number of known pathogens, which is very inconvenient. However, for all DNA or RNA present in the sample, mNGS allows detection of the entire microbiome and host genome or transcriptome in patient samples (Camargo et al., 2019; Wang et al., 2019; Irinyi et al., 2020; Li et al., 2020). Besides, the non-specificity of serum 1-3-β-D-glucan limits its application in the diagnosis of PJP (Del Corpo et al., 2020). 1-3-β-D-glucan is a common cell wall constituent of most pathogenic fungi, including P. jirovecii (Lu et al., 2011a; Li et al., 2015). The 1-3-β-D-glucan assay has been approved for making a diagnosis of invasive fungal disease (De Pauw et al., 2008). However, the role of the serum-1-3-β-D-glucan assay in the diagnosis of PJP is controversial, especially among patients with or without HIV infections (Li et al., 2015). Studies have demonstrated that a negative result of the serum-1-3-β-D-glucan determination is sufficient to rule out PJP in HIV cases only, but in non-HIV patients, 1-3-β-D-glucan assays are insensitive and nonspecific for invasive fungal disease (Li et al., 2015; Del Corpo et al., 2020), thus, the results should therefore be carefully analyzed in parallel with clinical features, radiological findings, and other diagnostic evidence (Lu et al., 2011a; Li et al., 2015; Del Corpo et al., 2020). In conclusion, these methods will lead to the delay of clinical treatment and affect the prognosis of patients. Therefore, it is necessary to find a method with high diagnostic accuracy, short detection time, and accurate identification of infectious pathogens (Jiang et al., 2021). mNGS has an efficient workflow, relatively low-cost consumption, and short turnaround time. Hence mNGS may be widely accepted in clinical practice (Han et al., 2019). Moreover, mNGS also has the characteristics of high throughput and high sensitivity and can measure millions or even hundreds of millions of nucleic acid sequences at the same time, which plays a very important role in the accurate diagnosis of infectious diseases (Yu et al., 2011; Wilson et al., 2014; Ni et al., 2015; Tong et al., 2015).The diagnostic performance of mNGS in the respiratory tract (Langelier et al., 2018; Xie et al., 2019), bloodstream (Dubourg and Raoult, 2016), central nervous system (Ramachandran and Wilson, 2020), pleural cavity (Chen et al., 2021), and prosthetic joint infections (Thoendel et al., 2018) has been appreciated. Research data also supports its advantages in detecting opportunistic pathogens and co-infections, especially uncultivable pathogens (Parize et al., 2017; Pan et al., 2019), such as Legionella pneumophila, Aspergillus spp. (Yue et al., 2021), Nocardia spp. (Ding et al., 2021), Mucor spp. (Liu et al., 2021) etc. In this study, clinical microbiology tests were compared with mNGS, and our data shows that the detection rate of mNGS for P. jirovecii in non-HIV infected PJP patients is significantly higher than that of serum 1-3-β-D-glucan and Methenamine silver staining, which can well auxiliary clinical diagnosis. Previous research shows that if the microorganism identified only by mNGS is accompanied by discrete evidence in clinical practice, the pathogen is considered a pathogen with a certain degree of clinical suspicion (Wu et al., 2020; Jiang et al., 2021). In our study, 34 patients with PJP were diagnosed through clinical comprehensive diagnosis and mNGS results, suggesting that mNGS plays an important role in assisting diagnosis of pathogens with a certain degree of clinical suspicion, which is consistent with the views in the above findings. We also analyzed the time interval from onset to the collection of samples and found that the interval time of the group with reads>100 was significantly shorter than that of the group with reads ≤ 100, which indicated that the number of reads detected by mNGS for P. jirovecii was closely related to the time of sample collection. This result is also consistent with previous studies, indicating that early treatment, effective use of antibiotics, and improved disease status will reduce the number of reads detected by mNGS for pathogens (Ai et al., 2018; Zhang Xx et al., 2019). In addition, we compared some factors that may affect the reads abundance of P. jirovecii detected by mNGS. The results of the univariate analysis showed that age, gender, white blood cells, hemoglobin, and platelets did not affect the reads abundance of P. jirovecii, while C-reactive protein, LDH, and procalcitonin were significantly different between the group with reads>100 and the group with reads ≤ 100.These results suggest that concurrent detection of inflammatory status during mNGS may determine the pathogen abundance, which is also consistent with previous studies (Zhang Xx et al., 2019). This study also has certain limitations. Firstly, due to the small sample size, the positive rate of clinical routine method evaluation is low, so the sensitivity of mNGS may be slightly overestimated which belongs to partial validation bias. Secondly, because BALF is adopted at different positions of lung segment, the partial deviation will occur, which will affect the reads abundance of P. jirovecii accompanied with the sensitivity evaluation. Thirdly, due to limited conditions, the hospital did not routinely perform PCR for P. jirovecii identification, so the diagnostic performance of mNGS and PCR was not compared. Finally, mNGShas disadvantages, for example, it is difficult for mNGS to distinguish P. jirovecii colonization from infection because there is no widely accepted mNGS quantification cutoff or threshold. In addition, the sequencing depth of mNGS is still limited, and the pathogen database needs further improvement (Gargis et al., 2016). Therefore, a definitive diagnosis of PJP must be based on a comprehensive summary of clinical features, laboratory abnormalities, imaging findings, and microbiological evidence, rather than mNGS alone (Jiang et al., 2021). However, the rapid development of next-generation sequencing technology will show higher sensitivity and specificity in diagnosing infections in the future (Thorburn et al., 2015), and we are confident that the mNGS will effectively give a valuable performance in clinical infection diagnosis.
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://bigd.big.ac.cn/gsa/browse/CRA008494 and the assigned accession of the submission is: CRA008494.
This study was reviewed and approved by Medical ethics committee of Sichuan Academy of Medical Sciences and Sichuan Provincial People’s Hospital. The patients/participants provided their written informed consent to participate in this study.
All authors contributed to the study conception and design. MZ, RY, LP, and MH wrote the manuscript draft. RY, XW, and ZG conducted most of the experiments. MZ and MH performed data analysis. LP and MH conceived the idea and directed the experiments. All authors contributed to the article and approved the submitted version.
This study was supported by Research and Development Projects of Science and Technology Department of Sichuan Province (Grant No.2019YFS0319), Sansure Biotech Transfusion Medicine Development Fund of Chinese Society of Blood Transfusion (CSBT-SX-2021-01) and Project of the Special Fund for Young and Middle aged Medical Research of the China International Medical Exchange Foundation(Z-2018-35-19202).
We acknowledge Dr. Fan Zhenxin (College of Life Sciences, Sichuan University) for bioinformatics helping, and Mr. Wang Baoshan (Mountain in Sight (Sichuan) Medical Technology Co. Ltd) for coordinating sample collection and shipping.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647190 | Cecilia Bañuelos,Abigail Betanzos,Rosario Javier-Reyna,Ausencio Galindo,Esther Orozco | Molecular interplays of the Entamoeba histolytica endosomal sorting complexes required for transport during phagocytosis | 27-10-2022 | cargo sorting,EhRabs,EhADH,endosome maturation,late endosome/MVB,intraluminal vesicles,LBPA,Vps | Entamoeba histolytica, the causative agent of human amoebiasis, exhibits a continuous membrane remodelling to exert its virulence properties. During this dynamic process, the Endosomal Sorting Complexes Required for Transport (ESCRT) machinery is a key player, particularly in phagocytosis, a virulence hallmark of this parasite. In addition to ESCRT, other molecules contribute to membrane remodelling, including the EhADH adhesin, EhRabs, actin, and the lysobisphosphatidic acid (LBPA). The endocytosis of a prey or molecules induces membrane invaginations, resulting in endosome and multivesicular bodies (MVBs) formation for cargo delivery into lysosomes. Alternatively, some proteins are recycled or secreted. Most of these pathways have been broadly characterized in other biological systems, but poorly described in protozoan parasites. Here, we encompass 10 years of ESCRT research in E. histolytica, highlighting the role of the ESCRT-I and ESCRT-III components and the EhADH and EhVps4-ATPase accessory proteins during phagocytosis. In particular, EhADH exhibits a multifunctional role along the endocytic pathway, from cargo recognition to endosome maturation and lysosomal degradation. Interestingly, the interaction of EhADH with EhVps32 seems to shape a concurrent route to the conventional one for MVBs biogenesis, that could optimize their formation. Furthermore, this adhesin is secreted, but its role in this event remains under study. Other components from the endosomal pathway, such as EhVps23 and LBPA, are also secreted. A proteomic approach performed here, using an anti-LBPA antibody, revealed that some proteins related to membrane trafficking, cellular transport, cytoskeleton dynamics, and transcriptional and translational functions are secreted and associated to LBPA. Altogether, the accumulated knowledge around the ESCRT machinery in E. histolytica, points it out as a dynamic platform facilitating the interaction of molecules participating in different cellular events. Seen as an integrated system, ESCRTs lead to a better understanding of E. histolytica phagocytosis. | Molecular interplays of the Entamoeba histolytica endosomal sorting complexes required for transport during phagocytosis
Entamoeba histolytica, the causative agent of human amoebiasis, exhibits a continuous membrane remodelling to exert its virulence properties. During this dynamic process, the Endosomal Sorting Complexes Required for Transport (ESCRT) machinery is a key player, particularly in phagocytosis, a virulence hallmark of this parasite. In addition to ESCRT, other molecules contribute to membrane remodelling, including the EhADH adhesin, EhRabs, actin, and the lysobisphosphatidic acid (LBPA). The endocytosis of a prey or molecules induces membrane invaginations, resulting in endosome and multivesicular bodies (MVBs) formation for cargo delivery into lysosomes. Alternatively, some proteins are recycled or secreted. Most of these pathways have been broadly characterized in other biological systems, but poorly described in protozoan parasites. Here, we encompass 10 years of ESCRT research in E. histolytica, highlighting the role of the ESCRT-I and ESCRT-III components and the EhADH and EhVps4-ATPase accessory proteins during phagocytosis. In particular, EhADH exhibits a multifunctional role along the endocytic pathway, from cargo recognition to endosome maturation and lysosomal degradation. Interestingly, the interaction of EhADH with EhVps32 seems to shape a concurrent route to the conventional one for MVBs biogenesis, that could optimize their formation. Furthermore, this adhesin is secreted, but its role in this event remains under study. Other components from the endosomal pathway, such as EhVps23 and LBPA, are also secreted. A proteomic approach performed here, using an anti-LBPA antibody, revealed that some proteins related to membrane trafficking, cellular transport, cytoskeleton dynamics, and transcriptional and translational functions are secreted and associated to LBPA. Altogether, the accumulated knowledge around the ESCRT machinery in E. histolytica, points it out as a dynamic platform facilitating the interaction of molecules participating in different cellular events. Seen as an integrated system, ESCRTs lead to a better understanding of E. histolytica phagocytosis.
The need for cell nourishment directs life activities from microorganisms to humans. To allow the uptake of nutrients, numerous molecules interact to perform diverse functions that involve distinct cell organelles. The plasma membrane (PM) acts as the interphase with the environment, where nutrients are present, and connects with the membranous system in the cytoplasm, where the ingested material is processed (Simpson, 2020). The nutrients uptake first depends on the interaction and turnover of receptors and ligands. Proteins forming channels and vesicles, and carrier proteins, act as vehicles to transport the prey (target cells) or cargo (molecules), or to biochemically transform it. The presence and quantity of factors involved in all these events are finely regulated in time and cellular location. Once ingested, the nutrients are degraded into simple compounds and, on the other hand, the cell recovers useful molecules for further utilization. Molecular cross-talking and membrane remodelling maintain the balance in all these processes and are the core of these events (Arlt et al., 2015; Palm and Thompson, 2017). Entamoeba histolytica, the parasite causative of human amoebiasis, is a professional devourer of bacteria and eukaryotic cells (Orozco et al., 1983). During capture, ingestion and digestion of the prey, the PM and internal membranes exhibit a particular dynamic activity, that, in addition to the movement of the prey inside the cell, transport the proteins that direct the events. When trophozoites invade human tissue, the nutrients are captured by endocytosis, phagocytosis and trogocytosis (Labruyère and Guillén, 2006; Nozaki and Nakada-Tsukui, 2006; Ralston et al., 2014). In response to an infection, both, the innate and adaptive immune responses are activated to control and eliminate invasive E. histolytica. Infiltrating trophozoites are attacked by cells of the complement system, and molecules present in the blood. Parasites are recognized by dendritic cells, which then activate CD4+ and CD8+ T cells, for developing a cellular immune response. CD4+ T cells produce IFN-γ, IL-4, and IL-5, whereas CD8+ T cells produce IL-17. IL-17 enhances secretion of IgA antibodies (humoral immune response) into the colonic lumen. IFN-γ stimulates macrophages to produce nitric oxide (NO) and neutrophils to release extracellular traps (NET). NO can directly kill amoebas, whereas NET can trap and immobilize them (Ávila et al., 2016; Díaz-Godínez et al., 2018; Uribe-Querol and Rosales, 2020). In turn, trophozoites use multiple pathogenic factors to resist the immune response and continue its survival and pathogenesis (Nakada-Tsukui and Nozaki, 2016). In the lumen of the large intestine, glycosidases and proteinases secreted from trophozoites, are involved in the degradation of the mucin mucus layer. The cysteine proteinase (EhCP)-A5 binds to and activates integrins on endothelial cells, leading to NLRP3 inflammasome formation. Also, the galactose and N-acetylgalactosamine (Gal/GalNAc) lectin binds to the Toll-like receptor 2, leading to NF-κB activation and release of inflammatory cytokines (IL-1, IL-6, IL-8, IL-12, IFN-γ and TNF-α) (Nakada-Tsukui and Nozaki, 2016; Uribe-Querol and Rosales, 2020). The EhCPADH complex, integrated by the EhADH adhesin and the EhCP112 cysteine protease (García-Rivera et al., 1999), and the prostaglandin E2, secreted from the amoeba, disrupt tight junctions of epithelial cells (Lejeune et al., 2011; Betanzos et al., 2013). Phagocytosis and trogocytosis are also involved in the removal of epithelial cells and invasion into the tissue (Labruyère and Guillén, 2006; Ralston et al., 2014). By their medical importance, their high activity during host invasion and the unique atypical organelles that the trophozoites have (Nozaki and Nakada-Tsukui, 2006; Smith and Guillen, 2010; Betanzos et al., 2019), they constitute an excellent system to study membrane remodelling, vesicular trafficking and their participation in the target cell attack. Our laboratory has studied molecules involved in phagocytosis. Data have shown that the Endosomal Sorting Complexes Required for Transport (ESCRT) machinery is involved in cellular functions, including phagocytosis, that require vesicle formation and membrane scission and repair. The ESCRT proteins interact with other molecules and together, they construct a chain of events that maintain the continuity of the process. Each one of the events and molecules of this chain are important, and if one of them is affected, the entire process is disturbed (López-Reyes et al., 2011; Avalos-Padilla et al., 2018; Galindo et al., 2022). In eukaryotes, the ESCRT machinery is formed by the complexes: ESCRT-0 (Vps27/Hrs, Hse1/STAM1), ESCRT-I (Vps23/Tsg101, Vps28, Vps37, Mvb12), ESCRT-II (Vps22/EAP30, Vps25/EAP20, Vps36/EAP45) and ESCRT-III (Vps20/CHMP6, Vps32/CHMP4, Vps24/CHMP3, Vps2/CHMP2) and the ESCRT accessory proteins (Bro1/Alix, Vps4-ATPase and Vta1/LIP5) (in parenthesis the names of the Saccharomyces cerevisiae/Homo sapiens proteins of each complex are indicated) ( Table 1 ) (Hurley, 2010; Schuh and Audhya, 2014). The majority of the genes and proteins of the ESCRT machinery are present in E. histolytica ( Table 1 ) (López-Reyes et al., 2011). ESCRT proteins change their cell location through phagocytosis, according to the advance of the event (López-Reyes et al., 2010). Their knock down or overexpression have repercussions in the rate of phagocytosis and in the in vitro and in vivo virulence expression (López-Reyes et al., 2010; Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018; Galindo et al., 2021; Galindo et al., 2022), highlighting their role in membrane transformation and in vesicles, tubes and tunnel-like structures formation, which in turn contribute to the transport of cargo, or the prey. To perform phagocytosis, the trophozoites need first to be attracted to and make contact with the prey (Orozco et al., 1985). Several ESCRT proteins associate to EhADH and Gal/GalNAc (Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018; Galindo et al., 2021), both involved in the adherence of trophozoites to the target cells (García-Rivera et al., 1999; Lopez-Vancell et al., 2000). Cellular signals initiate the pseudopodia formation and the capture of the prey. In this step, ESCRT proteins interact with actin, RabB and possibly other Rab proteins (Javier-Reyna et al., 2019). In general, Rab small GTPases play a fundamental role in signalling and activation of distinct molecules involved in the process (Saito-Nakano et al., 2005; Verma et al., 2020). ESCRT proteins have also been detected in phagosomes, late endosomes (LE), multivesicular bodies (MVBs) and their intraluminal vesicles (ILVs), in secreted vesicles, and in the MVBs fused to lysosomes (López-Reyes et al., 2010; Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018; Galindo et al., 2021; Galindo et al., 2022). These sequential events allow first, the transport of the prey through compartments containing different enzymes and molecules, ending in the prey digestion for recycling useful proteins, or in the secretion of others. EhADH acts as an adhesin and then, through the whole process, as a scaffold protein carrying molecules and interacting with distinct ESCRT proteins (Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018; Galindo et al., 2021; Galindo et al., 2022). Given the importance of endocytosis, particularly phagocytosis, and movement, in the E. histolytica virulence, we reviewed here the main known facts on membrane remodelling and protein transport during phagocytosis, with special emphasis in the role of the ESCRT machinery.
To perform the capturing of the prey, in general, trophozoites require to adhere to the target. The virulent strains of E. histolytica display a high rate of pinocytosis (micropinocytosis and macropinocytosis) and endocytosis (trogocytosis and phagocytosis) (Rodríguez and Orozco, 1986; Laughlin et al., 2004; Bettadapur and Ralston, 2020). Pinocytosis is the process by which the trophozoites absorb extracellular fluids and compounds (Laughlin et al., 2004), while in endocytosis, the cells capture macromolecules, cell surface components and even whole cells that include red blood cells (RBCs), live and apoptotic mammalian cells, and bacteria (Labruyère and Guillén, 2006). So far, there are no reports precising the mechanisms used by E. histolytica to discriminate among particles, molecules or whole cells to be internalized. Endocytic processes include phagocytosis, that refers to live whole, damaged or dead cells engulfment; and trogocytosis, where trophozoites ingest part of the living cells (Ralston et al., 2014; Bettadapur and Ralston, 2020; Nakada-Tsukui and Nozaki, 2021). The PM invagination and the formation of vesicles and vacuoles required for these processes, imply the renewal of the PM every 30 min (Doherty and McMahon, 2009). Particularly, phagocytosis begins with the adherence of trophozoites to target cells (Christy and Petri, 2011). The most characterized molecules in the primary contact are the Gal/GalNAc lectin (Singh et al., 2016) and the EhADH adhesin ( Figure 1 ) (García-Rivera et al., 1999). Other proteins involved in the contact to RBCs and human colorectal adenocarcinoma (Caco-2) cells, include the C2-domain–containing protein kinase (EhC2PK) (Babuta et al., 2020) and the lysine and glutamic acid rich protein (KERP1) (Seigneur et al., 2005), respectively. In the first step of phagocytosis, EhVps2, EhVps20, and EhVps32 proteins from the ESCRT-III complex (Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018) and EhVps23 from the ESCRT-I complex (Galindo et al., 2021) have been localized at PM, just in the place where the PM of trophozoites makes contact with the membrane of the target cell. During erythrophagocytosis, EhVps32 appears associated to the Gal/GalNAc lectin (Avalos-Padilla et al., 2015), while several proteins of ESCRT-I and ESCRT-III interact with the EhADH adhesin (Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018; Galindo et al., 2021) ( Figure 1 ). Future research will elucidate whether these ESCRT proteins help in some way to the adherence function. The Gal/GalNAc lectin is a heterodimer that comprises a transmembrane heavy subunit (Hgl, 170 kDa) and a glycosylphosphatidylinositol (GPI)-anchored light subunit (Lgl, 35/31 kDa), glycoproteins linked by disulphide bonds. The Hgl subunit contains a carbohydrate recognition domain that binds to D-galactose and N-acetyl-D-galactosamine and it is important for cell adhesion and extracellular matrix (ECM) degradation, and intervenes in both, colonization and contact-dependent cytotoxicity. Recent studies have suggested a role of Hgl in the ECM-mediated actin dot formation (Petri et al., 2002). On the other hand, EhADH is a multifunctional protein that contains at its N-terminal a Bro-1 domain, which makes it a member of the ALIX family (Bañuelos et al., 2005; Montaño et al., 2017). Its ability of binding to the target cell is conferred by an adherence domain present at the C-terminal (Arroyo and Orozco, 1987; García-Rivera et al., 1999). Monoclonal antibodies against the adherence domain inhibit trophozoite adhesion to and phagocytosis of erythrocytes (García-Rivera et al., 1999). In epithelial cells, this adhesin binds to proteins of the intercellular junctions, such as tight and adherens junctions and desmosomes, contributing to the epithelial damage and invasion (Betanzos et al., 2013; Hernández-Nava et al., 2017; Betanzos et al., 2018). These findings support the former evidences about the presence of trophozoites tightly adhered to the intercellular space between two epithelial cells, where these junctions are located (Martinez-Palomo et al., 1985). Of note, DNA vaccination of animals with the Ehadh and Ehcp112 genes, improves the immune response of hamsters against E. histolytica, and protects the animals from the damage caused in the liver by virulent amoeba strains (Martínez et al., 2009). In addition, mutant trophozoites with the Ehadh gene silenced show a reduction (30%) in their rate of erythrophagocytosis (Ocádiz-Ruiz et al., 2016). In contrast, EhADH overexpression increases (76%) the rate of phagocytosis by trophozoites (Bañuelos et al., 2012). EhADH is also an important accessory protein of the ESCRT machinery, and it frequently appears associated to other ESCRT members (Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018; Galindo et al., 2021). It binds to EhVps32 and EhVps23 through its Bro1 domain (Bañuelos et al., 2012; Galindo et al., 2021). Furthermore, the EhCPADH complex, the Gal/GalNAc lectin, and the EhVps32 (ESCRT-III) protein co-localize at the site of contact of trophozoites with RBCs (Avalos-Padilla et al., 2015). In this same attachment place, EhADH appears together with the EhVps23 (ESCRT-I) (Galindo et al., 2021), highlighting the role of the ESCRT machinery accompanying the Gal/GalNAc lectin and EhADH protein in their function as receptors. At the site of contact with RBCs, EhVps2 and EhVps24 are also present, suggesting that these ESCRT components also bind to surface proteins (Avalos-Padilla et al., 2018). Overall, these data indicate that several molecules participate to allow the trophozoites the specific contact with their prey. EhADH also associates to the cholesterol-trafficking proteins EhNPC1 and EhNPC2, suggesting an extra role for this molecule in the uptake and transport of this essential lipid toward cellular membranes (Bolaños et al., 2016). Although no relationship of cholesterol and the ESCRT machinery has been described yet, it is well known that cholesterol improves the adherence of trophozoites to the host cells and to the ECM (Mittal et al., 2008). On the other hand, several ESCRT proteins possess lipid-binding domains (Teo et al., 2006); thus, it is possible that these molecules act and interact with cholesterol during the membranes remodelling in phagocytosis. Membranes of trophozoites are mainly composed by phospholipids and cholesterol (Das et al., 2002). Particularly, the vesicle’s lipid and protein composition varies according to the function of the vesicle and its content (Goldston et al., 2012b; Castellanos-Castro et al., 2020), as shown using the model of giant unilamellar vesicles (GUVs) for reconstructing the ESCRT-III subunits assembly (Avalos-Padilla et al., 2018). To mimic the endosomal membranes composition, authors probed several lipids combinations, resulting phosphocholine:phosphoserine:cholesterol: phosphatidylinositol 3, phosphate (PI3P) in a 62:10:25:3 ratio, the optimal one (Avalos-Padilla et al., 2018). In addition to the transport of molecules during the vesicular trafficking, lipids also participate in the dynamic membrane fusion and fission. Several lipids have already been detected in the trophozoites during the capture and ingestion of the prey (Mittal et al., 2008; Welter et al., 2011; Castellanos-Castro et al., 2020). Phosphatidylinositol, a member of the family of glycerophospholipids, is phosphorylated at all combinations of D-3, 4, and 5 positions of the inositol ring, forming seven isotypes of phosphatidylinositol phosphates (PIPs). In eukaryotes, PIPs are localized at the PM and in membrane regions connecting them to the cytoskeleton. For instance, PIP2 is a critical regulator of actin polymerization and cytoskeleton/membrane linkages. The binding of cytoskeletal proteins to membrane PIP2 might alter lateral or transverse movement of lipids to affect raft formation or lipid asymmetry (Zhang et al., 2012). In E. histolytica, this lipid family is involved in phagocytosis and trogocytosis (Goldston et al., 2012a; Watanabe et al., 2020). PI(4,5)P2 is localized in the PM and mediates the signalling during cell adhesion (Goldston et al., 2012a). Nevertheless, it has not been experimentally probed whether the trophozoites lipids interact with the ESCRT machinery proteins since the first contact, as in other eukaryotes (Teo et al., 2006). The E. histolytica lipophosphoglycan (LPPG), mainly found in virulent strains, has been proposed as a molecule involved in the contact of the trophozoites and target cells (Stanley et al., 1992; Moody et al., 1998). However, so far, it has not been reported the interaction between the LPPG and subunits of the ESCRT machinery, or with other proteins involved in phagocytosis.
In eukaryotes, the receptor-ligand clustering produces changes in the PM composition and topology, triggering activation pathways (Zhdanov, 2017). In E. histolytica, the receptor-ligand clustering leads to the activation of the cytoskeleton, forming pseudopodia that surround the target, developing the phagocytic cup and participating in the active movement of the trophozoites (Verma and Datta, 2017). During the phagocytic cup formation, Ca2+ signalling (Jain et al., 2008) plays an important role in the recruitment of actin (Mansuri et al., 2014). There is a group of calcium-binding proteins (CaBPs), such as EhCaBP1 and EhCaBP3 known to regulate the dynamics of the cytoskeleton (Bhattacharya et al., 2006). These proteins interact with actin and phosphatidylserine that, along with actin remodelling and other molecules, produce the membrane deformation to form the phagocytic cup (Jain et al., 2008; Aslam et al., 2012). Other proteins involved in the organization and regulation of actin cytoskeleton during the cup formation are the small GTPases, such as the Rho, Rab, and Arf families (Bosch and Siderovski, 2013). EhRab21 and its effector EhC2B, which binds to phosphatidylserine in the presence of calcium, are localized in the advancing tips of the phagocytic cup, where EhC2B catalyses actin polymerization (Tripathi et al., 2020). EhRab35 is also present in the phagocytic cups ( Figure 2 ). In addition, the expression of an EhRab35 dominant negative in trophozoites, reduces the formation of phagocytic cups (Verma and Datta, 2017). By in silico analysis, it has been demonstrated that EhRabB binds to the EhADH Bro1 domain through its switch I zone; and to actin by the switch I and II regions (Javier-Reyna et al., 2019). These regions are crucial for the switching between GTP- and GDP-bound forms (Bos et al., 2007), facilitated by a guanine nucleotide exchange factor (Lee et al., 2009). In human, these states exhibit structural differences, allowing selective recognition of Rabs by regulatory and effectors proteins in a nucleotide-dependent manner (Pylypenko et al., 2018). In E. histolytica, by immunoprecipitation and immunofluorescence assays, it has been demonstrated that EhRabB binds to EhADH and actin, and all of them co-localize at the phagocytic cup ( Figure 2 ) (Javier-Reyna et al., 2019). In addition, EhVps23 from the ESCRT-I complex, and EhVp32, a member of ESCRT-III, are also localized at the phagocytic cups ( Figure 2 ) (Avalos-Padilla et al., 2015; Galindo et al., 2021). Thus, it is plausible to hypothesize that these proteins are directly or indirectly interacting with EhRabB and actin. The lipids such as PI3P and PI(3,4,5)P3 are localized at the phagocytic ( Figure 2 ) and trogocytic cups, where PI3P-binding proteins are recruited though FYVE domains (Powell et al., 2006; Nakada-Tsukui et al., 2009; Byekova et al., 2010). The E. histolytica genome is predicted to encode eleven FYVE domain-containing proteins (FPs) (Nakada-Tsukui et al., 2009). In yeast and human, the FYVE is a characteristic domain of the Vps27/HRS protein from the ESCRT-0 complex (Katzmann et al., 2003). In E. histolytica, the EhVps27 protein also harbors the FYVE domain (López-Reyes et al., 2011). Another FP protein, is EhFP4, which interacts with PIPs on the PM, thereby recruiting proteins such as EhRacC and EhRacD, both involved in the polymerization of actin filaments at the phagocytic cup (Nakada-Tsukui et al., 2009; Watanabe et al., 2020). These findings, point out to the participation of some lipids, small GTPases and its effectors, together with the actin cytoskeleton and some ESCRT members during the prey capture, including the phagocytic cup formation and the actin cytoskeleton organization.
Phagocytosis by professional phagocytes such as macrophages and immature dendritic cells in humans, and of course E. histolytica trophozoites, involves prey recognition, capture, and internalization by fusion of the extended pseudopodia. Once the prey has been internalized, it passes through a series of pleomorphic tubule-vesicular compartments, collectively called endosomes (Schuh and Audhya, 2014). In human and yeast, endosomes undergo maturation from EE to LE, which involves decreased luminal pH (6.0-4.9), a change in PIPs composition, and recruitment and activation of Rabs (Scott et al., 2014). In E. histolytica, the difference between EE and LE is not entirely clear, since the composition of PIPs (PI3P, PI[4,5]P2 and PI[3,4,5]P3) is similar in distinct stages of the endosome maturation, and there are no conclusive specific markers for each type of endosomes. In trophozoites, the phagosomes are formed once the phagocytic cup detaches from the PM (Nakada-Tsukui et al., 2009; Babuta et al., 2015). For the detachment, EhCaBP3 and EhCaBP5 activate myosin IB, which is involved in the closure of the pseudopodia for the formation and release of the phagosome (Aslam et al., 2012; Kumar et al., 2014). Although in human the dynamin is key for the liberation of the nascent vesicle, in E. histolytica, dynamin-like proteins have only been found in the nuclear membrane and mitosomes, but not in endosomes (Jain et al., 2010; Makiuchi et al., 2017). Despite the absence of dynamin in endosomes, other proteins such as ESCRT-III members contribute to the vesicle’s scission. In in vitro experiments, using the GUVs model, EhVps20, EhVps32, EhVps24 and EhVps2 participate in the formation and release of ILVs (Avalos-Padilla et al., 2018). Nevertheless, experiments using mutant trophozoites in the ESCRT components would demonstrate the participation of these proteins in vivo in the endosomes release. Alternatively, during 5 to 10 min of phagocytosis, pre-phagosomal vacuoles (PPV) are formed. These structures function as a temporary reservoir for digestive enzymes, transport amoebapores to the phagosome, and contain EhRab5 and EhRab7A (Saito-Nakano et al., 2004; Verma et al., 2015). Dissociation of EhRab5 from PPV promotes the fusion of this compartment with phagosomes of the endocytic pathway (Saito-Nakano et al., 2004). On the other hand, during the classic pathway for endosome maturation, the first structure formed is the EE and then, the LE or multivesicular bodies (MVBs) ( Figure 3 ). In human, Rab5 and Rab7 proteins are specific markers for EE and LE, respectively (Langemeyer et al., 2018). In E. histolytica, EhRab5 has also been suggested as a possible EE marker (Verma and Datta, 2019), while EhRab7A and EhRabB have been found in EE or LE (Saito-Nakano et al., 2007; Javier-Reyna et al., 2019; Verma and Datta, 2019) ( Figure 3 ). In addition, some proteins of the ESCRT machinery (EhVps2, EhVps4, EhVps20, EhVps23, EhVps24, EhVps32, and EhADH) have been localized in different endosomal structures during erythrophagocytosis (López-Reyes et al., 2010; Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018; Galindo et al., 2021) ( Figure 3 ). Since EE and LE specific markers are not yet available for E. histolytica, the accurate location of ESCRT proteins in the different structures cannot be fully determined yet. Nevertheless, based on the time of erythrophagocytosis, it can be suggested that EhVps23 and EhADH are present in EE (<5 min) (Galindo et al., 2021) ( Figure 3A ). Meanwhile, EhADH has been found in many endosomal vesicles during phagocytosis, from the first contact of the trophozoite with the target cell to the processes of digestion and secretion (Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Galindo et al., 2022). It suggests that this adhesin participates in the entire target cell capture and digestion process. In yeast and human, the ESCRT-III complex participates in the LE maturation (Teis et al., 2008). Accordingly, in E. histolytica EhVps2, EhVps4, EhVps20, EhVps24 and EhVps32, and EhADH, are found in LE (>15 min) ( Figure 3B ) (López-Reyes et al., 2010; Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018).
Some LE appear with several ILVs (~50 nm) inside and are referred as MVBs (100-600 nm in diameter), whose content is sent to lysosomal degradation. In other organisms, the ESCRT machinery mediates membrane budding and ILVs formation (Henne et al., 2011; Schuh and Audhya, 2014; Christ et al., 2017), and given the experimental data available, we postulate that in E. histolytica, ESCRT complexes and accessory proteins are pivotal for these events that form part of the phagocytosis phenomenon ( Figure 3B ). The MVBs formation initiates with the ESCRT-0 complex recruitment (Hurley, 2008). In E. histolytica, putative sequences of EhVps27 and EhHse1, members of this complex, were in silico identified (López-Reyes et al., 2010). However, these proteins lack the VHS (Vps27, Hrs and STAM1), and UIM (ubiquitin-interaction motif) domains, involved in lipid and ubiquitin binding, respectively (López-Reyes et al., 2011). Dictyostelium discoideum lacks the typical ESCRT-0 members, but Tom1 performs the ESCRT-0 activity (Blanc et al., 2009). Tom1 proteins harbor a GAT (GGA and Tom1) domain for ubiquitin (Ub) binding, and motifs for clathrin, membrane phospholipids, and Vps23 association (Blanc et al., 2009; Herman et al., 2011; Schuh and Audhya, 2014). In this context, E. histolytica has a Tom1 (EhTom1) homolog that, according to our preliminary analysis, possesses an Ub-binding domain, suggesting that it can perform functions similar to those assigned to orthologous members in other organisms. Next, ESCRT-0 recruits to ESCRT-I. The E. histolytica genome contains two genes (Ehvps23 and Ehvps37) of the ESCRT-I complex and their transcription increases after 5 min of erythrophagocytosis (López-Reyes et al., 2010), suggesting that a higher amount of these proteins is necessary for events related to phagocytosis when the MVBs are formed or being formed. The EhVps23 protein is located in MVBs and cytoplasmic vesicles in the basal state, while during phagocytosis it is found in phagosomes and vesicles close to the them (Galindo et al., 2021), strongly suggesting that it is participating in the molecular events performed at this stage of phagocytosis ( Figure 3B ). Moreover, the knock-down of the Ehvps23 gene causes a lower rate of phagocytosis (50% less), in contrast to trophozoites overexpressing EhVps23, which exhibit a higher rate of phagocytosis (20% more) than the control trophozoites (Galindo et al., 2021; Galindo et al., 2022). In addition, EhVps23 associates to EhADH, EhVps32, LBPA and EhUb, colocalizing in different endosomal structures. As EhADH, EhVps23 could play a role at various points of the endocytic pathway. Another experimental data on the relevance of EhVps23 in phagocytosis is the fact that trophozoites overexpressing EhVps23 migrate five-fold faster than control parasites, in concordance with the low rate of migration exhibited by Ehvps23-knocked down trophozoites (Galindo et al., 2022); this, points out to the participation of the ESCRT machinery also in motility, a fundamental event for phagocytosis. In regard to EhVps37, López-Reyes et al. (2010) found an EhVps37D protein lacking the alpha helixes required for ESCRT-I binding (Kostelansky et al., 2007; Galindo et al., 2021). Given the diversity exhibited by E. histolytica ESCRT proteins compared to their orthologues (Leung et al., 2008; López-Reyes et al., 2010; López-Reyes et al., 2011; Avalos-Padilla et al., 2018), the possibility of other unidentified ESCRT-I members, cannot be ruled out. In eukaryotes, after assembly of the ESCRT-I, the ESCRT-II complex is recruited to the endosomal membrane (Hurley, 2008). The E. histolytica genome contains the Ehvps22, Ehvps25 and Ehvps36 genes that form the ESCRT-II in this parasite (López-Reyes et al., 2011). López-Reyes et al. (2010) also found that Ehvps36 is downregulated after 5 min stimulation with RBCs. However, more studies at protein level are necessary to elucidate the participation of this complex during phagocytosis. Then, the ESCRT-II module recruits ESCRT-III proteins that remain in the cytoplasm in an inactive state, which is modified by the binding of Vps25 (ESCRT-II) to Vps20 (ESCRT-III) (Im et al., 2009). ESCRT-III proteins (Vps2, Vps20, Vps24 and Vps32) are regulated through an autoinhibitory switch mechanism that allows a tight control for their assembly. Vps20 suffers a conformational change to generate an open/close configuration that corresponds to an active/inactive state, respectively. Active Vps20 binds to the endosomal membrane, recruits the active Vps32, whereas Vps24 and Vps2 promote polymerization (Bajorek et al., 2009). In E. histolytica trophozoites, the four proteins are detected in erythrocytes-containing phagosomes of acidic nature, and in cytoplasmic vesicles at the PM proximity (Avalos-Padilla et al., 2018). ESCRT-III proteins also participate during the formation and release of ILVs within MVBs ( Figure 3B ). Electron microscopy images display EhVps32 forming the helical structures present on LE and phagolysosomes, necessary for endosomal membrane strangulation and ILVs release. The role of the ESCRT-III in ILVs formation was confirmed using the GUVs model. By this system, it was possible to rebuild in vitro the whole ESCRT-III machinery and establish the order of assembly of the proteins. First, the active EhVps20 binds to the GUVs membrane and recruits EhVps32, promoting membrane invaginations. EhVps24 allows the detachment of nascent vesicles, forming ILVs; and EhVps2 modulates their size (Avalos-Padilla et al., 2018). The key role of the ESCRT-III complex in phagocytosis is supported by the increased rate of erythrophagocytosis in parasites overexpressing EhVps32 (Avalos-Padilla et al., 2015). In contrast, when Ehvps20, Ehvps24 and Ehvps32 genes are silenced, a decrease in the erythrophagocytosis rate is observed (Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018). Besides to ESCRT-III proteins, EhVps23 (ESCRT-I) and EhADH (ESCRT accessory protein), probably also contribute to the membrane remodelling involved in the formation of ILVs ( Figure 3B ) (Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Avalos-Padilla et al., 2018; Galindo et al., 2021). In addition to EhADH, the ESCRT machinery needs the help of other accessory proteins, such as the EhVps4-ATPase and EhVta1 (López-Reyes et al., 2010). The EhVps4-ATPase encloses the MIT (Microtubule Interacting and Transport) domain at the N-terminal end, an AAA (ATPase Associated with a variety of Activities) motifs, and the C-terminal end. In human, Vps4-ATPase binds to CHMP2, CHMP4 and CHMP6 from the ESCRT-III, through its MIT domain, and it is probable that this also occurs in E. histolytica. EhVps4 has ATPase activity in vitro as in other eukaryotes, which depends on the conserved E211 residue (López-Reyes et al., 2010). The EhVps4-ATPase protein localizes around the phagocytosed erythrocytes ( Figure 3B ) when it is overexpressed, and accordingly, mutant trophozoites expressing EhVps4-E211Q exhibit a decrease in their rate of phagocytosis. In yeast and human, Vps4-ATPase enhances its ATPase activity due to Vta1 binding (López-Reyes et al., 2010). The activation of Vps4 by Vta1 implies Vps4 oligomerization and the enhancing of ATP hydrolysis by Vps4 oligomer (Azmi et al., 2006). Vta1 has VSL (Vps4, SBP1 and LIP5) and MIT domains, which allow its interaction with Vps4-ATPase and ESCRT-III proteins, respectively (Yeo et al., 2003). E. histolytica has the Ehvta1 gene, which is transcribed in basal conditions and phagocytosis (López-Reyes et al., 2010; López-Reyes et al., 2011). So far, bioinformatics analysis has not revealed the presence of VSL and MIT domains in EhVta1; however, we cannot discard interactions with EhVps4-ATPase and ESCRT-III members by other regions. Further research will unveil if the activity of EhVps4 depends on or is potentiated by EhVta1. On the other hand, in mammalian cells, an alternative ubiquitin-independent MVBs formation pathway, in which Alix and ESCRT-III proteins are involved (Dores et al., 2012), has been reported. In human, PAR1-activated receptor directly binds to Alix; then, Alix recruits CHMP4B and the rest of the ESCRT-III subunits, in an ubiquitin-independent manner (Dores et al., 2012). This also could be happening in E. histolytica, suggesting the participation of EhADH in MVBs formation, since it acts as an erythrocyte receptor by its adherence domain, whereas by its Bro1 domain, it recruits EhVps32. Overall, these data point out to the relevance of the concerted interactions among E. histolytica ESCRT proteins, as in other eukaryotes, for the membrane remodeling during the endocytic pathway ( Figure 3B ). In E. histolytica, some of these interactions need to be experimentally validated to confirm their role during phagocytosis.
The endosome maturation culminates with the phagosome fusion to the lysosome, forming the phagolysosome, which contains hydrolytic enzymes (EhCP1, EhCP2, EhCP4, EhCP5, amoebapores, phospholipases, dipeptidyl aminopeptidase, β-hexosaminidase, and lysozymes) (Long-Krug et al., 1985; Nickel et al., 1998; Flockenhaus et al., 2000; Que et al., 2002; Andrä et al., 2003; Furukawa et al., 2012); membrane receptors (Gal/GalNAc lectin) (Petri et al., 2002); and proteins related to cytoskeletal rearrangements and vesicular trafficking, such as EhRabs ( Figure 4 ) (Marion et al., 2005; Okada et al., 2006; Saito-Nakano et al., 2007; Verma and Datta, 2017). In this step, the role of Rabs is crucial to the protein transport and their degradation in lysosomes. Particularly, EhRab7B regulates EhCPs and holo-transferrin transport to lysosomes (Saito-Nakano et al., 2007; Smith and Guillen, 2010; Verma et al., 2015). Besides, the expression of an EhRab7B mutant reduces phagocytosis, lysosome acidification, and intracellular EhCPs activity (Saito-Nakano et al., 2007). The overexpression of EhRab35 increases the number of lysosomal compartments and the degradation of RBCs (Verma and Datta, 2017; Constantino-Jonapa et al., 2020). All these experimental results give strong evidence on the role of Rab proteins in different steps of phagocytosis. In human, the ESCRT machinery plays an additional role on endolysosomal organelles, responding to and promoting the repair of damaged or perforated membranes (Chen et al., 2019). Alix co-localizes with CHMP4A and other ESCRT-III components on endolysosomes after acute damage produced by L-leucyl-L-leucine O-methyl ester. The silencing of tsg101 and alix genes, from the ESCRT-I complex and ESCRT accessory proteins, respectively, prevents the recruitment of ESCRT-III members to the lysosomes, reducing the repair of these structures (Skowyra et al., 2018). Moreover, using live cell imaging, it was demonstrated that ESCRTs respond to small perforations in endolysosomal membranes and enable compartments to recover from limited damage (Radulovic et al., 2018; Skowyra et al., 2018). The need to protect the endolysosomal integrity has broad implications for many situations, perhaps most critically in highly phagocytic cells as E. histolytica, that internalizes and processes substantial loads of disruptive material. In this parasite, EhVps32 and EhADH as part of the ESCRT-III and ESCRT accessory proteins, respectively, as well as the molecules that bind to ESCRT components, such as EhRabB and LBPA, are present in the phagolysosomes during RBCs degradation ( Figure 4 ) (Bañuelos et al., 2012; Avalos-Padilla et al., 2015; Castellanos-Castro et al., 2016a; Javier-Reyna et al., 2019). Besides, the ESCRT-III proteins exhibit a propensity to assemble into spirals on highly curved membranes, and surround, constrict and ultimately close vesicles (Wollert et al., 2009; Henne et al., 2012; Avalos-Padilla et al., 2018). Thus, we hypothesize that the ESCRT-III proteins may repair damaged membranes on phagolysosomes, resealing wounds. Here, the EhADH protein, together with EhVps23, could be contributing to the recruitment of ESCRT-III components, since both bind to EhVps32.
After endocytosis, transmembrane cargo reaches endosomes, where it encounters complexes dedicated to opposing functions: degradation and recycling (Schuh and Audhya, 2014; Scott et al., 2014). There is scarce data about the molecules involved in the recycling pathway in E. histolytica and even less, regarding to the ESCRT’s role in this event ( Figure 5A ). By the way, Rab proteins are key membrane trafficking organizers that could be contributing to integrate and coordinate the ESCRT complexes towards the recycling pathway (Arlt et al., 2015). Rab8 and Rab11, together with their effector proteins, coordinate the control of proteins trafficking from the trans-Golgi to the PM in mammalian cells (Chen et al., 1998). In E. histolytica, the transport of receptors to the PM is necessary for binding to host cells. EhRab8A primarily resides in the endoplasmic reticulum (ER) and participates in phagocytosis. Its down-regulation by small antisense RNA-mediated transcriptional gene silencing remarkably reduces adherence and phagocytosis of erythrocytes, bacteria and carboxylated latex beads. Moreover, the surface expression of several proteins presumably involved in E. histolytica target recognition, is reduced in the EhRab8A in italics gene-silenced strain, indicating that EhRab8A regulates transport of surface receptors for the prey from the ER to the PM (Hanadate et al., 2016). EhRab11A translocates to the cell surface upon starvation, and it has been implicated in the transport of cyst wall components such as enolase, to the cell surface via actin filaments (Herrera-Martínez et al., 2013). EhRab11B is associated with non-acidified vesicles considered as recycling compartments, and regulates the secretion of EhCP1, EhCP2 and EhCP5 in a brefeldin insensitive manner (Mitra et al., 2007). EhRabB and the actin cytoskeleton participate in the transport of the EhCPADH complex towards the trophozoite PM (Javier-Reyna et al., 2019). As we have mentioned throughout this work, EhADH participates in several points of the endocytic pathway, where it has been localized with proteins of the ESCRT machinery. Likewise, the presence of some of ESCRT components close to the PM, such as EhVps2, EhVps20, EhVps23 and EhVps32, suggests that these proteins may participate in the mobilization of EhADH to the PM to interact with RBCs ( Figure 5 ). The role of ESCRT in cargo recycling, and its relationship with EhRabs have not been reported yet in E. histolytica. In Trypanosoma brucei, TbRab28 colocalizes with the ESCRT-I component Vps23 and is required for the turnover of internalized surface glycoproteins (Lumb et al., 2011). Furthermore, the T. brucei ubiquitylated invariant surface glycoprotein (ISG65) is rescued from lysosome delivery by the ESCRT accessory protein TbVps4-ATPase, to be recycled to the cell surface. In addition, the phosphoinositide-dependent binding of the ESCRT-III component TbVps24, affects the ISG65 traffic and accelerates its surface pool depletion. TbVps24 localizes to TbRab7 late endosome, and binds PI(3,5)P2 (Umaer and Bangs, 2020). Authors propose a model in which T. brucei ESCRT-III and ESCRT accessory components operate at two sites, one PI(3,5)P2 -dependent (degradation) and one PI(3,5)P2 -independent (recycling), to regulate ISG65 homeostasis. In this context, considering that late ESCRT proteins such as EhVps24 and EhVps4-ATPase are expressed by E. histolytica, we hypothesize that these proteins could exhibit a similar role during protein recycling in trophozoites.
Another alternative recycling pathway is driven by the retromer, a complex of proteins that recycle transmembrane receptors from endosomes to the trans-Golgi network (Seaman, 2012). The ESCRT and retromer pathways drive opposing endosomal functions that must somehow coexist; therefore, understanding the segregation of ESCRT and retromer domains on the endosome may be particularly relevant to understanding endosome function. In E. histolytica, the retromer is formed by EhVps26, EhVps29, EhVps35 (Loftus et al., 2005; Nakada-Tsukui et al., 2005), EhSNX1 and EhSNX (Batra et al., 2021), and has been associated with the EhCP’s pool maintenance (Nakada-Tsukui et al., 2005). The association among EhVps26, EhVps29, and EhVps35 was evidenced by immunoprecipitation and mass spectrometric analysis (Nakada-Tsukui et al., 2005). By immunoprecipitation experiments using α-EhVps23 antibodies and mass spectrometry analysis, our group detected EhVps26 and EhVps35 as interacting partners of EhVps23 (Galindo et al., 2022). In E. histolytica, the retromer associates with EhRab proteins. EhRab7A regulates the recycling of a non-identified EhCP receptor from the phagosomes to the trans-Golgi network (Saito-Nakano et al., 2004; Nakada-Tsukui et al., 2005; Nozaki and Nakada-Tsukui, 2006). EhRab7A binds to a sequence rich in charged amino acids located at the C-terminal end of EhVps26. EhRab7A overexpression produces the reduction of EhCPs activity, but EhVps26 overexpression restores it (Nakada-Tsukui et al., 2005; Nozaki and Nakada-Tsukui, 2006). Instead, the EhVps29 overexpression leads to a reduction of intracellular EhCPs activity (Srivastava et al., 2017). Altogether, these data suggest a role for the retromer as a machinery essential for the restitution of EhCP receptors, as it has been demonstrated in other eukaryotes (Wang et al., 2018), most likely via protein trafficking. Despite the relative absence of evidences around the relationship among the retromer and the ESCRT machinery, possible regulatory interactions should exist to balance degradative and recycling functions, as it has been reported for Caenorhabditis elegans, in which segregating microdomains are enriched in the retromer from those enriched in ESCRT-0 for maintaining the required balance between recycling and degradation activities within the endosome pathway (Norris et al., 2017). Since endosomes are a mosaic of functional arrays consisting of transmembrane cargo, lipids, and peripheral membrane proteins, it is expected that these components segregate and exhibit a dynamic performance in trophozoites, changing over space and time. Thus, as an endosome matures, it will change in structure and function, determining the final fate of cargo molecules. Overall, findings suggest that recycling pathways are cross-regulated by ESCRT to maintain an appropriate balance of each activity within a given endosome and its sorting. Further research will confirm the role of the ESCRT machinery in proteins recycling in E. histolytica.
During its pathogenic mechanism, E. histolytica secretes several molecules to reach the target cell and initiate invasion or we can speculate that also to communicate with other trophozoites. Likewise, cytolysis of target cells is carried out by cytolytic proteins secreted to the extracellular medium, including EhCPs (Que et al., 2002) and amoebapores (Zhang et al., 2004). Hydrolytic enzymes secretion is led by a specific interaction ligand-receptor, conducting to a dynamic vesicular transport and cytoskeletal rearrangement ( Figure 5A ) (Nozaki and Nakada-Tsukui, 2006). Other molecules that have been experimentally probed as secreted products by this parasite are: prostaglandin E2, EhNPC1 and EhNPC2, and some ESCRT components, including EhVps23 and EhADH ( Figure 5A ) (García-Rivera et al., 1999; Ocádiz et al., 2005; Sato et al., 2006; Lejeune et al., 2011; Bolaños et al., 2016; Galindo et al., 2022). Particularly, ESCRT components are secreted in vesicles (Galindo et al., 2022), which probably carry molecules that participate in the prey capture or in cell-cell communication ( Figure 5A ). Regarding to this, in mammalian cells, ILVs from MVBs also modulate intercellular communication when they are targeted to the PM for being secreted as exosomes. Here, an Alix- and ESCRT-III–dependent pathway promotes the sorting and delivery of tetraspanins to exosomes (Larios et al., 2020). However, in extracellular vesicles (ECVs) secreted by E. histolytica, tetraspanins have not been found yet (Sharma et al., 2020), although EhADH is there ( Figure 5A ) (Galindo et al., 2022). In addition, as mentioned above, EhADH provides an additional pathway for MVBs formation (Avalos-Padilla et al., 2015), and, by this route, this ESCRT accessory protein could control the targeting of exosomal proteins. The ESCRT-III proteins recruitment to the endosomes could occur independently of other ESCRTs (Avalos-Padilla et al., 2018), but might require LBPA, as it has been reported in human and yeast (Larios et al., 2020). The ESCRT-related proteins that have been found interacting with LBPA in E. histolytica, include EhVps23 and EhADH (Castellanos-Castro et al., 2016b; Castellanos-Castro et al., 2016a; Galindo et al., 2021). By a proteomic approach, 496 secreted proteins from cultured trophozoites were immunoprecipitated using an α-LBPA antibody, followed by mass spectrometry analysis (LC-ESI-HDMSE). After theoretical astringent conditions, only 221 proteins exhibited high confidence values. The graph ( Figure 6 ) shows that the highest percentages of proteins obtained correspond to the non-identified category (27%), followed by metabolite interconversion enzymes (24%), translational proteins (15%), cytoskeletal proteins (10%), and proteins involved in membrane trafficking (9%). From these groups, we enlist 44 proteins related to endocytosis, vesicular trafficking and cytoskeleton ( Table 2 ), and the rest of the proteins in the Supplementary Table I . Next, we performed a gene ontology (GO) enrichment analysis to the 221 secreted proteins immunoprecipitated with α-LBPA, to gain insights into the cellular functions and biological processes related to them ( Figure S1 ). Regarding the cellular component terms, the predominant categories correspond to membrane-related and cytoskeleton proteins; meanwhile, the molecular function terms referred to binding, cytoskeleton, and catalytic activity. The analysis for the biological processes in which proteins could be involved, pointed out that presumably, they are participating in metabolic events, translation and cytoskeleton organization ( Figure S1 ). In conclusion, this analysis evidences that most of the secreted proteins interacting with LBPA are mainly cytoskeletal proteins or related to them, revealing that cytoskeleton results fundamental for the transport and secretion of proteins. Of note, the assays were carried out in basal conditions and possibly for this reason, in the proteome, EhADH and EhVps23 were not detected. Also, it is feasible that the amount of proteins or their high susceptibility to degradation, derived in undetectable products. Further experiments under erythrophagocytosis conditions should be performed to confirm the interaction of these ESCRT proteins with LBPA.
In this review, we recapitulated the known interactions between distinct ESCRT complexes’ subunits, and virulence factors and molecules from other nature in the E. histolytica protozoan. Some approaches used at this moment to study the ESCRT machinery include the bioinformatical analysis of its members, proteins modelling, molecular docking, the generation and phenotypic characterization of trophozoite mutants (summarized in Table 3 ), the in vitro reconstruction of GUVs, and importantly, the identification of new partners of ESCRT components by proteomic approaches. However, there is a challenge to extend this knowledge to the interactions required for the successful attack of the amoeba, as a biological entity that reaches molecules, cells and organs from the host. During host colonization, the uptake of cargo or the prey is central for trophozoites survival, and depends on endocytosis. This process demands a dynamic membrane remodelling and an active transport of molecules, which is finely regulated for the concerted performance of proteins. In this event, endosomes formation allows the sorting and internalization of a wide variety of molecules, and even whole cells, with the inherent biogenesis and trafficking of vesicles. Here, several molecules such as EhADH, EhRabs, actin and LBPA, participate. The interaction among these molecules inside and outside of trophozoites, have been revealed by both, bioinformatics and experimental strategies, where proteomics has elucidated meaningful data. In this review, we report proteins that are secreted to the extracellular environment and interact with LBPA, most of them related to endocytosis, membrane trafficking and cytoskeleton. Moreover, these molecules could act in an orchestrated fashion with the assistance of cellular machineries (i.e., ESCRT), resulting in a functional integrated system. The ESCRT machinery comprises an evolutionary conserved group of specialized proteins that have shed light on the mechanisms underlying protein sorting, MVBs biogenesis, membrane trafficking and cell signalling. Therefore, in this work, we put forward the putative role of the ESCRT system as a platform contributing for membrane remodelling, molecular transport to different compartments, endosome maturation, recycling and secretion. Remarkably, the ESCRT accessory protein EhADH emerges as a scaffold protein that assists the ESCRT machinery functions for connecting molecules and events along the endocytic pathway. Hence, E. histolytica is a suitable model of study for pleomorphism and phagocytic capacities, which results very useful for a comprehensive understanding of the biomedical implications of amoebiasis and its potential strategies of control.
CB, AB, RJ-R, AG, and EO contributed to the conception and design of the review. RJ-R performed the proteomic analysis. AG designed the table and figures. All authors contributed to manuscript revision, and read and approved the submitted version.
This work was supported by the National Council for Science and Technology (CONACYT) of Mexico (grants: A1-S8380 for EO and 284477 for AB), and RJ-R received a CONACYT Postdoctoral Fellowship.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The handling editor declared a shared affiliation with the authors at the time of review.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647198 | Chao Fang,Lin Zhou,Hui Huang,Hai Tong Xu,Tao Hong,Su Yue Zheng | Bioinformatics analysis and validation of the critical genes associated with adamantinomatous craniopharyngioma | 27-10-2022 | craniopharyngioma,pathways in cancer,CDH1,SHH,WNT5A | Adamantinomatous craniopharyngioma (ACP) is an epithelial tumor that arises when Rathke’s pouch remains during embryonic development. The pathogenesis of ACP remains unclear, and treatment options are limited. Here, we reveal the critical genes expressed in ACP and provide a basis for further research and treatment. The raw dataset GSE94349 was downloaded from the GEO database. We selected 24 ACP and 27 matched samples from individuals with no documented tumor complications (control group). Then, we screened for differentially expressed genes (DEGs) to identify key signaling pathways and associated DEGs. A total of 470 DEGs were identified (251 upregulated and 219 downregulated). Hierarchical clustering showed that the DEGs could precisely distinguish the ACP group from the control group (CG). Gene Ontology (GO) enrichment analysis indicated that the upregulated DEGs were mainly involved in cell adhesion, inflammatory responses, and extracellular matrix management. The downregulated DEGs were primarily involved in cell junction and nervous system development. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis indicated that the critical pathway was pathways in cancer. In the PPI network, CDH1, SHH, and WNT5A had the highest degrees of interaction and were associated with the formation of ACP. CDH1 was verified as a critical gene by quantitative reverse transcription–polymerase chain reaction (qRT-PCR) in ACP and CG samples. We found that CDH1 may play an important role in the pathways in cancer signaling pathway that regulates ACP development. The CDH1 gene may be a target for future research and treatment of ACP. | Bioinformatics analysis and validation of the critical genes associated with adamantinomatous craniopharyngioma
Adamantinomatous craniopharyngioma (ACP) is an epithelial tumor that arises when Rathke’s pouch remains during embryonic development. The pathogenesis of ACP remains unclear, and treatment options are limited. Here, we reveal the critical genes expressed in ACP and provide a basis for further research and treatment. The raw dataset GSE94349 was downloaded from the GEO database. We selected 24 ACP and 27 matched samples from individuals with no documented tumor complications (control group). Then, we screened for differentially expressed genes (DEGs) to identify key signaling pathways and associated DEGs. A total of 470 DEGs were identified (251 upregulated and 219 downregulated). Hierarchical clustering showed that the DEGs could precisely distinguish the ACP group from the control group (CG). Gene Ontology (GO) enrichment analysis indicated that the upregulated DEGs were mainly involved in cell adhesion, inflammatory responses, and extracellular matrix management. The downregulated DEGs were primarily involved in cell junction and nervous system development. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis indicated that the critical pathway was pathways in cancer. In the PPI network, CDH1, SHH, and WNT5A had the highest degrees of interaction and were associated with the formation of ACP. CDH1 was verified as a critical gene by quantitative reverse transcription–polymerase chain reaction (qRT-PCR) in ACP and CG samples. We found that CDH1 may play an important role in the pathways in cancer signaling pathway that regulates ACP development. The CDH1 gene may be a target for future research and treatment of ACP.
Craniopharyngioma (CP) is a complex and diverse congenital intracranial tumor, and surgical resection is the primary treatment strategy at present. Because this tumor is located near essential brain structures, such as the optic nerve and hypothalamus, it poses significant challenges for surgery. Damage to the hypothalamus and other important brain tissue structures can lead to high fever, electrolyte disorder, obesity, and other severe complications and seriously harm patients’ quality of life after surgery (1). Treatment options for CPs are difficult to determine because the balance of risks and benefits of surgery varies significantly from patient to patient. The close, heterogeneous relationship to the hypothalamus makes surgical removal of CPs challenging even though this remains the primary treatment strategy (1, 2). The incidence of CPs is 4.6% of all intracranial tumors, and its exact pathogenesis remains unclear (1). There are two theories regarding the origin of CP cells: embryogenetic theory and metaplastic theory (3). There are two histological types of CPs, ACP and papillary craniopharyngioma (PCP), of which ACP is the most common (4). As PCP is rare in clinical practice, all subjects in this study had ACP. Exome sequencing studies have demonstrated that PCP and ACP have distinct genetic origins, primarily driven by mutually exclusive alterations; BRAFV600E is observed in 95% of PCPs, and CTNNB1 is observed in 75%–96% of ACPs (5). For PCPs with BRAFV600E mutations, targeted therapy with BRAF inhibitors combined with MEK inhibitors can reduce tumor volume by 85% to 91% (6, 7). However, no similar clinical studies have been conducted on ACPs. In addition to CTNNB1, CDH1 and SHH may play essential roles in the occurrence of ACP (8, 9). Gene expression microarrays can be used to identify differentially expressed genes (DEGs) in ACP tissues, providing high-throughput gene expression data. Gene expression microarrays are an essential approach for studying the pathogenesis of CP. The National Center of Biotechnology Information GEO database comprises a large amount of genetic data, providing a rich resource for the study and analysis of differential gene expression levels (10). Zou et al. (11) analyzed CP data in the USA, revealing that the mechanisms of ACP occurrence and development might involve the regulation of the RNA polymerase II promoter and glutamate receptor binding. Li et al. (12) also analyzed sample data and identified MMP12 as a potential therapeutic target in ACP. The present study examined the GSE94349 dataset, comprising a large number of samples obtained from American patients (13). GSE94349, submitted by Donson et al. on 31 January 2017 (13), contains genetic data from 168 samples, analyzed by microarray technology using the Affymetrix Human Genome U133 Plus 2.0 array GPL570 platform. We selected 24 ACP surgical tumor samples and 27 samples from individuals with no documented tumor complications (control group, CG) from GSE94349. However, as Donson et al. (13) was a clinical trial, the researchers did not quickly achieve a perfect match between the two groups. Although these two studies have implications for ACP, our study further contributed GSEA data. Finally, we used the intersection of the Database for Annotation, Visualization, and Integrated Discovery (DAVID) and GSEA to determine the key genes. No relevant article has been published regarding the critical genes involved in ACP by using the intersection of DAVID and GSEA. By analyzing the gene expression data of GSE94349, DAVID and GSEA were used to conduct GO and KEGG pathway enrichment analyses. Then, the intersection of the two methods was obtained, namely, crucial pathways and genes. Finally, STRING and Cytoscape software were used to construct protein−protein interaction (PPI) networks of critical pathways and genes. The present study identified critical genes in the pathogenesis of CP, which will provide new insight into the treatment of CP.
The gene expression microarray data of GSE94349_series_matrix.txt were downloaded from the GEO (http://www.ncbi.nlm.nih.gov/geo/) database. The quality control and standardization of these data were completed prior to the current study. The data were processed using R software (version3.6; http://www.r-project.org/). First, we searched and downloaded GPL570, the corresponding platform of GSE94349, from the official GEO website. GSE94349 was matched with GPL570 in R language to complete gene ID conversion and annotation. Next, the missing values were calculated by KNN in R language to supplement the missing values.
DEGs were filtered by using the limma package in R (http://limma.org) (14). Only genes with | log2-fold change (Fc) | > 4 and p-values < 0.01 were considered DEGs. Finally, we obtained upregulated and downregulated DEGs. Volcano mapping was performed in R language; heatmaps were drawn using the R package gplot 2.
The upregulated and downregulated DEGs were input into the online tool DAVID (DAVID 6.8 version) for GO enrichment analysis. The results were analyzed according to biological process (BP), cellular component (CC), and molecular function (MF), and p < 0.01 was considered statistically significant.
DEGs were imported into the DAVID online tool for KEGG pathway enrichment analysis, and p < 0.01 was considered statistically significant. We imported all the gene data for the GSEA. The Kolmogorov−Smirnov test was used to calculate the value of DEGs in each KEGG pathway by 1,000 repeated permutation tests to conduct the KEGG pathway enrichment analysis of DEGs in GSEA. p < 0.01 was considered statistically significant. Then, the upregulated and downregulated DEG KEGG pathway enrichment analysis results were derived. Finally, the intersection of the results of the two analysis methods was obtained to determine the target DEGs. DAVID was used to conduct KEGG pathway enrichment analysis for all the different genes, yielding more comprehensive results. However, GSEA is a KEGG pathway enrichment analysis of all genes, yielding a complete dataset. The intersection of the two may be more valuable for further research (15). We used Venn diagrams to show the common KEGG pathways after the intersection of the DAVID and GSEA results.
The KEGG pathway-related DEGs from the intersection of the DAVID and GSEA results were imported into the Search Tool for the Retrieval of Interacting Genes (STRING). The gene interaction relationship was derived with a confidence score >0.7 as the cutoff standard. Then, Cytoscape software (Version 3.5.1) was used to construct the PPI network between the KEGG pathway and its related DEGs. CytoHubba was used to predict the top 10 key genes. Three critical genes (CDH1, WNT5A, and SHH) were selected.
ACP and CG samples were provided by the Department of Neurosurgery at The First Affiliated Hospital of Nanchang University (Nanchang, China). ACP samples were collected after endoscopic nasal resection, stored at 4°C for transport and then preserved in liquid nitrogen. Samples from individuals with no documented tumor complications [control group (CG)] were collected from patients undergoing surgery for epilepsy. The mean age of the four epileptic patients was 20 years (age range, 8–34 years), including two men and two women. A total of four ACP samples from two men and two women were obtained. The mean age of the ACP patients was 25.5 years (age range, 8–35 years). Patients pathologically diagnosed with ACP were included in the present study, while patients with other diagnoses were excluded. All specimens were pathologically and clinically diagnosed as ACP by three pathologists. The present study was approved by the Research Ethics Committee of Nanchang University [Nanchang, China; First Affiliated Hospital of Nanchang University (2020) Medical Research Ethics Review (No. 160)], and written informed consent was provided by all patients prior to the start of the study.
1. Clinical diagnosis of craniopharyngioma and consent to surgery; 2. Han nationality, Chinese, right-handed; 3. No history of chronic cardiovascular diseases; no history of infectious diseases; previously healthy, no history of kidney disease and liver disease; 4. Normal hearing and language functions, no metal implants in the body, no history of surgery for heart, lung, or other vital organs, no history of major diseases such as brain disorders, and no history of mental illness; and 5. At least two pathologists diagnosed ACP. 6. Only when the above five items meet the requirements can they be included in the sample database of this study.
1. Clinically diagnosed epilepsy and consent to surgery; 2. Han nationality, Chinese, right-handed; 3. No history of chronic cardiovascular diseases or infectious diseases; 4. No history of kidney and liver diseases; 5. Normal hearing, no metal implants in the body; and 6. No history of mental illness. Only when the above six items meet the requirements can they be included in the sample database of this study.
1. Oral hormone medication within 2 weeks; 2. Speech and hearing impairment; 3. Patients who cannot cooperate with researchers in information collection (e.g., coma patients with severe illness); 4. Lost contact or refused to answer the phone during the follow-up period so that postoperative information could not be collected; and 5. Patients with anemia, cachexia, and other blood cells or low hemoglobin. 6. Any one of the above five items should be excluded from this study.
1. Patients or their family members voluntarily withdraw from the study after surgery; 2. Non-craniopharyngioma patients were diagnosed after preoperative collection of craniopharyngioma patient specimens; 3. Postoperative complications that require hormone and other related treatments; 4. Postoperative complications and inability to cooperate with follow-up investigators; and 5. Patients who cannot cooperate with postoperative follow-up. 6. If any of the above five criteria are met, the study will automatically stop by default.
Three critical genes (CDH1, WNT5A, and SHH) were validated in ACP tissues and compared with CG tissues by qRT-PCR using the QuantStudio 7 Flex real-time PCR system (Bio-Rad, Nanchang, China). The primer sequences used are shown in Table 1 . The 2−ΔΔCt method was used, and the PCR results were normalized to the ACTIN gene (16). Total RNA was extracted from ACP samples according to the instructions of the RNA Extraction Kit (Servicebio, China). Then, the total RNA of ACP samples was reverse transcribed according to the steps of the Servicebio®RT First Strand cDNA Synthesis Kit (Servicebio, China). The total RNA and cDNA of CG samples were obtained by the same method. Finally, 7.5 μl of 2×qPCR Mix, 1.5 μl of 2.5 μm primer (upstream + downstream), 2.0 μl of reverse transcription product (cDNA), and 4.0 μl of ddH2O were added to a 200-µl PCR tube. After the PCR solution was prepared, it was gently mixed by pipetting up and down. A 96-well PCR plate was placed into a special bracket. The researchers avoided touching the bottom of the reaction plate to avoid affecting the data reading. Three wells were prepared for each reaction, and qPCR was conducted using the following reaction conditions: predenaturation at 95°C for 10 min, denaturation at 94°C for 15 s, and annealing at 60°C for 30 s for 40 cycles. When the temperature rose from 65°C to 95°C, the fluorescence signal was collected every 0.5°C to form a melting curve. RT-PCR (Bio-Rad, China) was performed on CG samples and ACP samples according to the above steps. Each experiment was repeated three times.
The ΔΔCT method was used as follows: A = CT (target gene, sample to be tested) − CT (internal marker gene, sample to be tested); B = CT (target gene, control sample) − CT (internal standard gene, control sample); K = A − B; express multiple = 2−K. All experimental data are expressed as the mean ± SD and were statistically analyzed by t-test. p < 0.05 was considered to be statistically significant.
The gene expression microarray data of GSE94349_series_matrix.txt were downloaded from GEO, comprising 168 samples. The ACP and control groups were selected from GSE94349_series_matrix.txt, and other samples were removed. Quality control and standardization of these data were completed, and all chip data reached comparable levels.
Twenty-four ACP samples and 27 control samples in the GSE94349 dataset were analyzed. The volcano plot in Figure 1 shows all genetic differences with a threshold of -log10 p > 2 (p < 0.01) and |log2 Fc| > 4. Red dots represent upregulated genes, and green dots represent downregulated genes. Then, based on the cutoff criteria (p < 0.01 and |log2 Fc| > 4), a total of 470 DEGs were identified, including 251 upregulated and 219 downregulated DEGs. All DEGs were analyzed. DEGs were also divided into upregulated and downregulated components for separate analyses. This study aimed to identify the upregulation and downregulation of target DEGs; thus, we divided the upregulated and downregulated genes for analysis. The expression heatmap of the DEGs is shown in Figure 2 . Hierarchical cluster analysis accurately distinguished ACP samples from control samples.
The DEGs were imported, DAVID was used for functional enrichment analysis, and the results were obtained according to the three functions of MF, CC, and BP. The GO analysis results of upregulated and downregulated DEGs are listed in the top 10 items in order of p-value in Figure 3 . Figure 3A shows the detailed results of the MF analysis of DEGs. The upregulated DEGs were mainly related to the activation of structural molecules and the binding of calcium ions; downregulated DEGs were mainly associated with calcium ions and calmodulin binding. Figure 3B shows the detailed results of the CC analysis of DEGs. The upregulated DEGs were mainly concentrated in exosomes, extracellular regions, and extracellular spaces; downregulated DEGs were mainly enriched in cell junctions, neuron projections, and neuronal cell bodies. Figure 3C shows the detailed results of the BP analysis of DEGs. The upregulated DEGs were mainly associated with cell adhesion, inflammatory responses, and extracellular matrix organization, and the downregulated DEGs were mainly associated with chemical synaptic transmission and nervous system development. The results showed that the upregulated genes were mainly involved in cell adhesion, inflammatory responses, and extracellular matrix management. The downregulated genes were mainly involved in cell junction and nervous system development.
Figure 4A shows the results of the DEG KEGG pathway analysis by DAVID. Seven KEGG pathways were enriched in the upregulated DEGs (p < 0.01). Twelve KEGG pathways were enriched among the downregulated DEGs (p < 0.01). The upregulated pathway comprised pathways in cancer, the PI3K-Akt signaling pathway, the extracellular matrix (ECM)–receptor interaction, and the Wnt signaling pathway. The downregulated pathways were mainly associated with pathways related to retrograde endocannabinoid signaling, the synaptic vesicle cycle, morphine addiction, and neuroactive ligand–receptor interactions. Figure 4B and Table 2 show the results of DEG analyses obtained by GSEA KEGG pathway analysis. Using p < 0.01 as the standard, there were seven upregulated and no downregulated pathways. Detailed results of the 7 KEGG pathways are shown in Table 2 . Upregulated pathways were mainly associated with pathways in cancer and cytokine−cytokine receptor interactions. The KEGG pathway analyses identified 19 significant pathways from DAVID and seven significant pathways from GSEA. The results of the two methods intersected, and only one common upregulated pathway was found, as shown in the Venn diagram in Figure 4C . The only upregulated pathway was “pathways in cancer”. The relationship between this unique pathway and related pathways and genes is shown in Figure 5 .
All of the 15 DEGs in the one common pathway were collected using STRING, and a PPI analysis was performed. The 15 DEGs were imported into the STRING database, and their interactions were identified. If the interaction score was >0.7, the PPI network could be formed with 14 nodes and 32 edges. The PPI networks presented in Figure 6 show that all DEGs were upregulated. As shown in Figure 7 , the cytoHubba plug-in was used in Cytoscape software to search for the top 10 key genes among the 15 DEGs in the PPI network map. Finally, the top three critical genes (CDH1, WNT5A, and SHH) were selected from the 10 genes for RT-PCR verification.
Quantitative reverse-transcription polymerase chain reaction (qRT-PCR) was used to measure and compare the expression levels of the key genes in the ACP and control groups ( Figure 8 ). The expression levels of three key genes were significantly increased in ACP vs. the control group, including CDH1, WNT5A, and SHH, but only CDH1 had statistical significance (p < 0.05) ( Figure 8 ).
A total of 470 DEGs were identified in the present study, including 251 upregulated and 219 downregulated DEGs. GO enrichment analysis indicated that the upregulated genes were mainly located in the extracellular exosome, extracellular region, and extracellular space and were involved in structural molecule activation and the binding of calcium ions, cell adhesion, inflammatory responses, and extracellular matrix organization bioprocesses. Another study showed that proinflammatory mediators drive the phosphorylation and activation of STAT3 (17). Persistent signaling through this pathway in cancer can result in a chronic inflammatory phenotype and suppression of antitumor immunity (13). Other studies have shown that cholesterol crystals in ACP activate the inflammasome, leading to the secretion of inflammatory cytokines that drive the inflammatory response (18). In addition, ECM proteins mediate epithelial–mesenchymal transformation (EMT) in ACP cells (19). The downregulated genes were mainly involved in cell junction and nervous system development. Research has shown that epithelial cell adhesion molecule (Ep-CAM) expression in craniopharyngioma could be a predictive marker of relapse (20). Additionally, ACP cells originate from remnants of Rathke’s cleft, which is neuroepithelial (1, 21). The KEGG pathway enrichment analysis in this study showed that “pathways in cancer” (https://www.kegg.jp/pathway/hsa05200) was the major pathway involved in ACP. There was only one common pathway and all 15 DEGs in this pathway were collected using STRING, and a PPI analysis was performed. In the DAVID KEGG analysis, these 15 DEGs were mainly involved in the following five signaling pathways: Wnt signaling pathway, ECM–receptor interaction, amoebiasis, PI3K-Akt signaling pathway, and pathways in cancer. Among these signaling pathways, the association between amoebiasis and ACP was not previously reported. The other four signaling pathways have been associated with ACP. The Wnt signaling pathway is an important aspect of ACP pathogenesis. CTNNB1-Mut has been shown to promote ACP primary cell proliferation by activating Wnt/β-catenin signaling (22). On the other hand, reducing the expression of β-catenin can significantly inhibit the proliferation of ACP cells, and β-catenin can promote the expression of Fascin mRNA and Fascin by acting on Fascin genes in the nucleus, thus promoting the migration and invasion ability of ACP cells (23). The ECM can mediate EMT in ACP cells (19). EMT plays an important role in ACP development. ECM can facilitate the migration and differentiation of cells and trigger EMT, which is important for the progression and metastasis of various tumors (24, 25). Overexpression of ECM in ACP cells can promote tumor proliferation, migration, and invasion (19). Although the association between the PI3K-Akt signaling pathway and ACP has not been reported, Andoniadou et al. showed that fibroblast growth factors (FGFs) could induce β-catenin cells to actively divide (26). As shown in Figure 5 , three FGF members (FGF4, FGF19, and FGF20) of the 15 DEGs were involved in the PI3K-Akt signaling pathway. The FGF family proteins are key regulators of several biological processes, and together with their receptors, they affect the development of many human cancers (27). Overactivation mutations in FGF receptors have been identified in several human cancers, including breast, bladder, and prostate cancers (28, 29). It is tempting to speculate that FGFs induce active division of β-catenin cells through the PI3K-Akt signaling pathway. However, three FGF members of the 15 DEGs were also involved in pathways in cancer; therefore, FGFs induce active division of β-catenin cells through the PI3K-Akt signaling pathway or pathways in cancer, which requires confirmation in further experiments. Similarly, the association between pathways in cancer and ACPs has not been reported, but the three most significant key genes (CDH1, WNT5A, and SHH) have been closely/positively associated with craniopharyngioma. One report showed that SHH is highly expressed in ACPs (9). After the formation of Rathke’s sac, the expression of SHH in this region gradually decreases (30). SHH can promote β-catenin overexpression in ACP animal and cell models by paracrine and autocrine effects, respectively (26). In addition, studies have shown that the expression of VEGF in ACP cells can promote tumor angiogenesis and increase microvascular density (31); however, whether the SHH pathway affects tumor blood supply in ACP through VEGF has not been clarified. This result provides new directions for future research on ACPs. As a sequestering protein of β-catenin, E-cadherin (CDH1) plays an important role in canonical Wnt signaling (32). Previous studies have reported conflicting results regarding the expression of E-cadherin, encoded by the CDH1 gene, in ACP. Preda et al. found that CDH1 expression in ACPs was not correlated with β-catenin (33). However, Qi et al. found that β-catenin was positively correlated with CDH1 expression in ACP; β-catenin might regulate CDH1 expression in ACPs, and decreased CDH1 expression in ACPs has been associated with tumor recurrence (34). However, Barros et al. found that CDH1 expression in ACPs was not associated with tumor recurrence (32). All these findings need to be further verified. However, our results provide a new direction for the future treatment of ACP. Cancer-associated fibroblasts (CAFs) are a major component of the cancer stroma and promote cancer cell aggressiveness by secreting multiple factors (35). WNT5A was found to be highly expressed in tumor stromal fibroblast gastric cancer studies and was associated with poor prognosis (36). Wnt/β-Catenin and WNT5A/Ca pathways regulate proliferation and apoptosis of keratinocytes in psoriasis lesions (37). It is tempting to speculate that the Wnt/β-Catenin and WNT5A/Ca pathways regulate the proliferation and apoptosis of cells, representing a potential new therapeutic target for treating ACP in the future. However, no similar study has been performed for ACP. Although the key genes revealed by RT-PCR in this study are relatively clear, the sample size was small. A larger number of samples are needed for validation. The roles of the three key genes remain unclear and require verification in future studies. WNT5A could be used as a target for the treatment of ACP in the future. In conclusion, we hope that these methods can be applied in future studies. DAVID was used to thoroughly analyze the DEGs, while GSEA was performed to comprehensively analyze the genes. The main aim of our study was to analyze the KEGG pathways of all genes, including the DEGs, using GSEA. The analysis of large amounts of ACP gene data requires substantial work, and the analysis results are likely to change as the major databases are updated. Our results suggest that the 15 DEGs, including CDH1, WNT5A, and SHH genes and “pathways in cancer”, may be associated with ACP. According to the cytoHubba statistical results in Cytoscape software, the first three DEGs were selected for RT-PCR verification. Our results suggest that CDH1 may play an important role in the pathways in cancer signaling pathway that regulate ACP development. However, its specific role in ACP remains to be confirmed in further experiments. Brastianos et al. showed that in PCPs with the BRAFV600E mutation, a BRAF inhibitor combined with MEK inhibitor targeted therapy can reduce tumor volume by 85%–91% (6, 7). We expect that ACP can be eliminated by similar treatments in the future. The limitation of this study is that most of the conclusions were drawn from bioinformatics analyses, and previous experimental results are lacking in in-depth research. Clinical craniopharyngioma is rare. Our neurosurgery department treats approximately 12 patients with craniopharyngioma annually, among which 2–3 have PCP and 1–2 decline surgical treatment. Therefore, it is difficult to obtain a large number of samples for validation in a short time period. In addition, at present, methods for craniopharyngioma cell culture are in the early stages; thus, the characteristics of these tumors are challenging to verify in cell experiments. In addition, no other analytical methods were used in this study; thus, more reliable conclusions may be obtained by integrating other analytical methods, such as ingenuity pathway analysis and WebGestalt analysis. For the three key genes selected in this study, we hope to further verify these three critical genes identified in this study and explore their specific mechanisms of action in ACP. In particular, we expect that CDH1 will be studied in the future to determine how it regulates the occurrence and development of ACP through the pathways in cancer signaling pathway. According to the results of this study, we concluded that pathways in cancer signaling pathway is an important signaling pathway in the development of ACP. CDH1, WNT5A, and SHH regulate ACP formation through this pathway. Although this study concludes that CDH1 is the most critical gene, due to the small sample size of this study, we will expand the sample size in the future to further verify this result. For the study of ACP, in addition to CTNNB1 mutation expressing β-catenin to activate the classical Wnt signaling pathway and promote the proliferation and migration of tumor cells, current studies have confirmed that SHH in ACP can also participate in the regulation of the Wnt signaling pathway by promoting the expression of β-catenin through autocrine or paracrine. At the same time, β-catenin can regulate the expression of EGFR and promote the migration of tumor cells (38–40). Unfortunately, these studies have only been carried out in animals and cells and have not been confirmed in clinical trials. In addition, due to the lack of stable ACP cell lines, although many researchers have extracted primary ACP cells from fresh ACP tissues, no stable ACP cell lines have been constructed, so the results of these cell experiments are difficult to accept by the public.
Publicly available datasets were analyzed in this study. This data can be found here: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE94349.
This study was reviewed and approved by The Research Ethics Committee of Nanchang University [NanChang, China; First Affiliated Hospital of Nanchang University (2020) Medical Research Ethics Review (No. 160)]. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.
TH and SZ conceived the study and supervised the research. CF performed the bioinformatics analysis and was a major contributor to the writing of the manuscript. LZ prepared the figures and edited the manuscript. HH performed the PCR validation experiments. All authors contributed to the article and approved the submitted version.
This study was supported by the National Natural Science Foundation of China (grant no. 82060246), the Natural Science Foundation of Jiangxi Province, and the Science and Technology Research Project of Jiangxi Provincial Education Department (grant nos. 20202BABL206059 and GJJ190128).
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647202 | Swapna Thomas,Maria K. Smatti,Allal Ouhtit,Farhan S. Cyprian,Muna A. Almaslamani,Asmaa Al Thani,Hadi M. Yassine | Antibody-Dependent Enhancement (ADE) and the role of complement system in disease pathogenesis | 10-11-2022 | Coronaviruses,ADE,Complements,C1q,Non-neutralizing antibodies,COVID-19 | Antibody-dependent enhancement (ADE) has been associated with severe disease outcomes in several viral infections, including respiratory infections. In vitro and in vivo studies showed that antibody-response to SARS-CoV and MERS-CoV could exacerbate infection via ADE. Recently in SARS CoV-2, the in vitro studies and structural analysis shows a risk of disease severity via ADE. This phenomenon is partially attributed to non-neutralizing antibodies or antibodies at sub-neutralizing levels. These antibodies result in antigen-antibody complexes' deposition and propagation of a chronic inflammatory process that destroys affected tissues. Further, antigen-antibody complexes may enhance the internalization of the virus into cells through the Fc gamma receptor (FcγR) and lead to further virus replication. Thus, ADE occur via two mechanisms; 1. Antibody mediated replication and 2. Enhanced immune activation. Antibody-mediated effector functions are mainly driven by complement activation, and the first complement in the cascade is complement 1q (C1q) which binds to the virus-antibody complex. Reports say that deficiency in circulating plasma levels of C1q, an independent predictor of mortality in high-risk patients, including diabetes, is associated with severe viral infections. Complement mediated ADE is reported in several viral infections such as dengue, West Nile virus, measles, RSV, Human immunodeficiency virus (HIV), and Ebola virus. This review discusses ADE in viral infections and the in vitro evidence of ADE in coronaviruses. We outline the mechanisms of ADE, emphasizing the role of complements, especially C1q in the outcome of the enhanced disease. | Antibody-Dependent Enhancement (ADE) and the role of complement system in disease pathogenesis
Antibody-dependent enhancement (ADE) has been associated with severe disease outcomes in several viral infections, including respiratory infections. In vitro and in vivo studies showed that antibody-response to SARS-CoV and MERS-CoV could exacerbate infection via ADE. Recently in SARS CoV-2, the in vitro studies and structural analysis shows a risk of disease severity via ADE. This phenomenon is partially attributed to non-neutralizing antibodies or antibodies at sub-neutralizing levels. These antibodies result in antigen-antibody complexes' deposition and propagation of a chronic inflammatory process that destroys affected tissues. Further, antigen-antibody complexes may enhance the internalization of the virus into cells through the Fc gamma receptor (FcγR) and lead to further virus replication. Thus, ADE occur via two mechanisms; 1. Antibody mediated replication and 2. Enhanced immune activation. Antibody-mediated effector functions are mainly driven by complement activation, and the first complement in the cascade is complement 1q (C1q) which binds to the virus-antibody complex. Reports say that deficiency in circulating plasma levels of C1q, an independent predictor of mortality in high-risk patients, including diabetes, is associated with severe viral infections. Complement mediated ADE is reported in several viral infections such as dengue, West Nile virus, measles, RSV, Human immunodeficiency virus (HIV), and Ebola virus. This review discusses ADE in viral infections and the in vitro evidence of ADE in coronaviruses. We outline the mechanisms of ADE, emphasizing the role of complements, especially C1q in the outcome of the enhanced disease.
Antibodies induced by infection and vaccination can be a double-edged sword, as they play a vital role in protection, however in certain cases can enhance the illness. Such differential effects of antibody response depend on many factors, including the targeted epitope on the virus, cross-reactivity with host proteins, glycosylation pattern of the antibody-Fc fragment, host complement system, and others (Borsos and Rapp, 1965, Shim, 2011). In part, the virus may utilize the non-neutralizing antibodies bound to viral surface proteins for a more efficient entry into target cells and thus, elevates the viral infection (Hohdatsu et al., 1991). This phenomenon of increased viral infectivity by sub-neutralizing concentrations of antibodies or by non-neutralizing antibodies is termed antibody-dependent enhancement (ADE). The interrelation of prior available antibodies with the increased severity of disease progression has been perceived in many respiratory viruses, including RSV, measles (Kim et al., 1969, Nader et al., 1968), and other viruses including Flaviviruses (Peiris and Porterfield, 1982), Human immune deficiency virus (HIV) (Robinson et al., 1988a), and Ebola virus (EBOV) (Takada et al., 2003, Takada et al., 2001). In vitro studies also showed evidence of ADE in SARS, MERS, and COVID-19 (Iankov et al., 2006, Osiowy et al., 1994, Wan et al., 2020, Wu et al., 2020, Yip et al., 2016). However, in respiratory infections, the non-neutralizing antibodies might lead to an immune complex formation that could be deposited in the lung or other tissues, causing complement deposition, enhanced inflammation and immunopathology (Nader et al., 1968, Graham, 2016a). This review will focus on the role of complements in ADE.
ADE can occur via two different mechanisms: antibody-mediated replication and enhanced immune activation ( Fig. 1). The antibody-mediated replication is mainly observed in viruses that infect immune cells, including Dengue and HIV (Dejnirattisai et al., 2010, Gorlani and Forthal, 2013), where the virus enters the cell via Fc-FcR (Fc on the antibody and FcR on cells) and further replicates inside the cells. This is otherwise called extrinsic ADE. Extrinsic ADE occurs when the virus, in the presence of sub-neutralizing levels or non-neutralizing antibodies, infects FcR expressing cells, including macrophages or monocytes (Dejnirattisai et al., 2010). The FcR is a receptor expressed predominantly on the surface of immune cells and possesses a vital role in the immune system's protective functions. The FcR interacts with the Fc portion of the antibody when the Fab portions bind to the antigen surface resulting in virus- immune complex entry in to cells (Mancardi and Daëron, 2014). There are three main classes of FcRs; Fc gamma receptor (FcγR), Fc alpha receptor (FcαR), and Fc epsilon receptor (FcεR). The FcR involved in the ADE is FcγR (Mancardi and Daëron, 2014). On the other hand, the enhanced immune activation involves the formation of antigen-antibody-complement formation and deposition in certain tissues, particularly, respiratory system. This type of ADE mechanism is observed in non-macrophage tropic viruses, primarily respiratory viruses, including RSV and measles (Kim et al., 1969, Nader et al., 1968, Graham, 2016b). Though this mechanism occurs due to the non-neutralizing antibodies however, the disease enhancement is mediated via excess secretion of pro-inflammatory cytokines and complement deposition in the tissues. This is otherwise called intrinsic ADE (Nader et al., 1968, Polack et al., 2002a). The complement cascade is composed of more than 50 small plasma proteins and glycoproteins synthesized primarily by liver and also by tissue macrophages and monocytes. These proteins and glycoproteins function as a cascade to help immune system eliminate the virus by inducing series of inflammatory responses (Byrne and Talarico, 2021, Dunkelberger and Song, 2010). Complement activity is consumed or activated by antigen-antibody complex. The first complement to get activated is C1q and further C3 to C9 complements gets activated, and convertases are formed to release the final membrane attack complex (MAC). The MAC attacks and destroys the infected cell along with the virus-antibody complex (Dunkelberger and Song, 2010). Complement activation has been reported to be associated with disease severity in dengue (Churdboonchart et al., 1983) and HIV (Füst et al., 1994) infections. Moreover, complement activation and deposition are reported in respiratory infections including RSV-induced infection after formalin-inactivated RSV(FI-RSV) vaccination followed by RSV challenge in mice (Melendi et al., 2007). The infected mice weeks after vaccination were reported to enhance complement activation compared to infected mice without any vaccination. Recent studies have observed enhanced complement activation and deposition in patients with a severe infection in COVID-19 patients (de Nooijer et al., 2021, Gao et al., 2020a, Zinellu and Mangoni, 2021). For coronavirus, a non-macrophage tropic virus, the ADE mechanism would involve the intrinsic (complement-mediated) mechanism by activating complement and cytokine pathways leading to obstruction in the airway tissues. A recent study reports the association of elevated serum C3a with the disease severity and mortality in severe COVID-19 patients (Henry et al., 2021a, Henry et al., 2020). More studies are required to understand the role of complements in disease enhancement and the exact mechanism in ADE. This would spread light to more efficient vaccine development strategies.
The first report on ADE was made in 1964 by Hawkes et al (Hawkes, 1964). The study assessed arbovirus neutralization using antiviral antibodies against four different flavivirus strains. They observed that virus-specific antibodies, especially immunoglobulin (Ig) G, at sub-neutralizing concentrations, elevated viral titers in chick embryonic cells (Hawkes, 1964, Hawkes and Lafferty, 1967). However, at high titers of antibodies, the infection was prevented. Thenceforth, ADE has been identified in many virus-cell systems, predominantly for the Flaviviridae family. However, dengue viruses (DENV) are the most studied, including four serotypes (Halstead et al., 1970, WHO, May,2021). The emergence of severe or fatal dengue shock syndrome (DSS) and dengue hemorrhagic fever (DHF) during the 1960 s in Thailand was investigated by Halstead et al. and associates. Experimental studies have shown that individuals who have a pre-existing dengue immunity are more likely to develop severe dengue shock syndrome (DSS) and dengue hemorrhagic fever (DHF) (Halstead et al., 1970, Halstead et al., 1967). A long-term study (2004–2016) conducted among children between 2 and 14 years observed a significant relationship between dengue disease severity and pre-existing antibodies to dengue serotypes. The results showed a 7.64-fold higher hazard of DHS or DSS in children with antibodies from previous infection (Katzelnick et al., 2017). Similarly, mice that received West Nile Virus (WNV) and DENV antisera from infected blood donors demonstrated IgG-mediated ADE when challenged with zika virus (ZIKAV) (Bardina et al., 2017). The mice that received DENV positive plasma exhibited a 21.4% survival rate only, while those that received control plasma showed a 93.3% survival rate. In the same study, the in vitro analysis using K562 cells, DENV, and WNV antisera enhanced the infection of K562 cells when treated with ZIKAV. However, the infection was reduced when treated with IgG-depleted sera (Bardina et al., 2017). Several experimental studies reported that the FcγR on the immune cells mediates ADE (Porterfield, 1986, Sondermann et al., 2001). The interaction of the FcγR with the Fc on the antibody is observed in experimental studies using sera from secondary infection (ADE) (Halstead, 1977). Sera from mice immunized with Zaire Ebola Virus (ZEBOV) glycoprotein was found to induce enhanced infectivity of VSV pseudotyped Ebola virus glycoprotein (EBOV GP) via FcR-mediated mechanism in human kidney cells (293 cell line) (Takada et al., 2001). Furthermore, in convalescent human plasma from EBOV-infected individuals, VSV pseudotyped Ebola virus glycoprotein (EBOV GP) showed enhanced internalization into 293 cells (Takada and Kawaoka, 2003). In 2020, C1q mediated mechanism of ADE was reported in EBOV infection (Furuyama et al., 2020). The study reported that the mechanism of disease pathogenesis in EBOV is mediated via the cross-linking of C1q receptor on the immune cell surface with virus-antibody complex. ADE was also reported in HIV infections via FcR and complement-mediated mechanisms (Robinson et al., 1987). Human monoclonal antibodies directed to the transmembrane glycoprotein 41, such as V10–9, N2–4, and 120–16, increased HIV infectivity in vitro in MT2 (human cord leukocyte) cells. According to Robinson et al., T cells that express complement receptor 2 (CR2) are more likely to enhance HIV infectivity, implying that CR2 and CD4 play a vital role in complement-mediated disease enhancement (Robinson et al., 1988b). Mitchell et al. also reported the role of CR2 in ADE during HIV infection. They reported that murine monoclonal antibodies to CR2 and CD4 reduced HIV infection in vitro (Robinson et al., 1990). Similarly, another study reported the role of C1q in enhancing the neutralization of WNV by humanized IgG1 and IgG3 (Mehlhop et al., 2009a). ADE was also reported with other viral families. In the 1980 s, sub-neutralizing antibody titers against Sindbis viruses and Semliki Forest were reported to enhance the infection of macrophages in vitro experiments (Chanas et al., 1982). Enhanced infection of macrophages with Ross Rover virus (RRV) was also shown to be FcR mediated in the presence of diluted antisera from infected patients (Linn et al., 1996). ADE was reported in animals also. FcR mediated ADE mechanism is reported in cats immunized against feline infectious peritonitis virus (FIPV). Cats were passively induced with FIPV serum antibodies. These cats had poor survival rates when infected with FIPV compared to the control ones. The study demonstrated ADE by entering a non-neutralizing antibody-virus complex into the peritoneal and alveolar macrophages leading to enhanced infection and disease outcomes (Weiss and Scott, 1981). The study also reported a faster decline of antibody titers in FIPV infected kittens when compared to non-infected controls. The mechanism of this internalization of the virus was later studied in vitro ADE assay via FcγRIIa on the immune cells (Takano et al., 2008). A recent study reported the role of C1q in ADE during Ebola virus infection via FcγR independent mechanism in human kidney cells (Furuyama et al., 2020). The study suggested that virus-anitbody-C1q complex can enter the cells via C1q receptor leading to enhanced replication. Another study in 2012 reported the role of Mannose binding lectin (MBL) in dengue virus infection (Shresta, 2012). The study reported an association between depressed level or activity of MBL protein with disease severity in dengue infection. Extensive complement mediated ADE was reported by HIV-1 strain in presence of non-neutralizing autologous antibodies was reported in the year 2011 by Neli et al. They observed enhanced infection of HIV in T cells expressing CR2 (Willey et al., 2011). These shreds of evidence indicate that non-neutralizing antibodies facilitate immune cell infection via FcR- mediated virus entry. However, in non-macrophage tropic viruses, including respiratory viruses, the disease enhancement mediated by pre-existing antibodies from infection and vaccination is reported in many viruses.
ADE can result in Enhanced respiratory disease (ERD), leading to acute lung injury and severe clinical symptoms (Acosta et al., 2015). The characteristics of ERD include monocytic infiltration and surfeit of eosinophils in the respiratory tract (Polack, 2007). This phenomenon can occur due to homotypic or heterotypic infection by a different serotype after natural infection, vaccination, or by the transfer of passive maternal immunity (Su et al., 2021). ADE in respiratory infections has been mostly reported to be vaccine-associated disease enhancement (VADE) and the mechanism involved is intrinsic in most of the respiratory viruses.
The first study report on ADE in respiratory infection was in 1969, following RSV vaccination trial. Formalin-inactivated RSV (FI-RSV) vaccination resulted in a higher incidence of increased hospitalization due to severe illness in children (80%) when compared to non-immunized (5%) (Kim et al., 1969). Since then, many in vitro studies have shown increased viral infectivity of different cell lines in the presence of vaccinated sera. In vitro enhancement of RSV infection in macrophages (U937 cells) was reported by Gimenez et al. when treated the cell lines with the virus in the presence of diluted human serum samples from RSV infected children (Gimenez et al., 1989). Another study reported the enhanced infection of macrophages when monoclonal antibodies are directed to F (Fusion glycoprotein) and G (attachment glycoprotein) surface glycoproteins on RSV, which are the targets for neutralizing antibodies (Ananaba and Anderson, 1991, Krilov et al., 1989a). However, this virus uptake was much lesser when either Fc on the antibody or FcR on the cells were blocked (Krilov et al., 1989a). Further, Gimenez et al.,1996, reported their observation of neutralizing and enhancing abilities of the RSV antibodies with the help of four monoclonal antibodies (MABs) against G and F glycoproteins (Gimenez et al., 1996a). All four MABs showed significant ADE in U937 (human macrophage) cell line. They observed the same when a mixture of two MABs (against G and F) was used, which indicates that these antibody responses are synergic. Another two studies reported that immunization of mice with FI-RSV elicited a T-helper cell type 2 (Th2) dominant response, which enhances ADE (Moghaddam et al., 2006, Castilow et al., 2007). In challenged mice with RSV, the enhanced disease along with lung inflammation and injury was found to be associated with pulmonary eosinophilia. This is linked to the Th-2 cytokine response by CD4 T cells (Castilow et al., 2007). Such Th-2 response was recorded for both F glycoprotein subunit and inactivated virus, resulting in the formation of immune complexes in the lungs of infected mice (Ananaba and Anderson, 1991). Gomez et al. reported the infection of lung dendritic cells via antibody-mediated virus (RSV) uptake that affected the normal T-cell activation (Gómez et al., 2016). In a set of in vitro experiments, Wicht et al. demonstrated an increased RSV infection in human monocytes (THP-1) in the presence of sub-neutralizing concentrations of RSV antibodies (van Erp et al., 2019). Similarly, antibodies produced against the FI-RSV vaccine caused ADE in cotton rats (van Erp et al., 2017). Van et al. and associates also reported in their findings that human maternal antibodies also enhance viral infection in FcR-bearing human cell lines in in vitro experiments as the antibodies were less neutralizing. Hence the association between reduced virus neutralization and ADE is clear in severe RSV infection (van Erp et al., 2017). On the other hand, the enhanced pathology by weakly neutralizing antibodies through forming immune complexes and complement activation and deposition were observed in two children who died of enhanced RSV infection (Polack et al., 2002a). The formation of immune complexes and complements was demonstrated in mice experiments by immunizing the mice with FI-RSV and then challenging them with RSV (Polack et al., 2002a). In 2003, Simos et al. studied ADE in Bonnet monkeys. They observed increased RSV infection in FI-RSV immunized monkeys when compared to non-immunized counterparts (Ponnuraj et al., 2003). However, the clinical effects of ADE in RSV infection are still not thoroughly described.
The first observation of enhanced virus replication concerning influenza infection was studied in a rat model in 1980. The study aimed to evaluate the response elicited after immunization using a subunit vaccine against heterologous challenges with different subtypes of Influenza A virus (Askonas and Webster, 1980). Another study showed enhanced infection of macrophage-like cell lines (P388D1) by Influenza A subtype H1 NWS virus and two other antigenic drift strains. Enhanced infection was observed in the presence of cross-reactive antibodies against Influenza A viruses through FcγR-dependent mechanisms (Ochiai et al., 1990). Another study reported that pigs immunized with two different inactivated swine influenza viruses (H1N1 and H1N2) vaccines exhibited protection against homologous infection however, increased infectivity was seen in a heterologous infection module (Tamura et al., 1991). The study demonstrated antibodies produced against hemagglutinin (HA) and neuraminidase (NA) influenza virus-promoted virus uptake by antigen-presenting cells (APCs). The enhanced infection was FC-mediated but not Fab-mediated. A significant increase in viral infection was observed in the presence of sub-neutralizing concentrations of antibodies to the homologous virus (Tamura et al., 1991). Another study reported the role of cross-reactive anti H1N1 HA antibodies that were associated with the enhanced disease. This study characterized sera from pigs immunized with whole inactivated H1N1 vaccine, and the titers of neutralizing antibodies against the H1N1 virus were high. However, Fc mediated disease enhancement was observed in pigs when challenged with the H1N1 virus (Khurana et al., 2013). Another recent study reported the ADE of influenza virus in a heterologous challenge. Infection caused enhanced lung infection in piglets when treated using maternally derived antibodies from patient’s sera (Rajao et al., 2016). A recent study in mice reported ADE during H3N2 infection in mouse model. Mice receiving lowest doses of monoclonal antibodies (MAb 78/2) against H3N2 strain experienced lung homogenate analysis. The results showed high levels of proinflammatory cytokines as well as several Th2 cytokines (Winarski et al., 2019a). Different studies comparing the effect of influenza virus uptake in the presence of sera from vaccinated or naturally infected individuals indicated that antibodies from both natural infection and attenuated vaccine enhanced the uptake of different strains in the heterologous challenge. This denotes the Fc-FcR mediated cellular entry and complement mechanism of ADE in the influenza virus.
Coronaviruses belong to a large family, Coronaviridae, that infects a wide range of species and causes various diseases (Masters, 2006). Seven different strains of human coronaviruses have been identified: four viruses that cause the common cold ( 229E, NL63, OC43, and HKU1) and three viruses (SARS-CoV, MERS-CoV, and SARS-CoV-2) that cause respiratory infections (Woo et al., 2010). Coronaviruses possess spikes on their surface that facilitate their attachment and entry into host cells (Huang et al., 2020). In both SARS-CoV and SARS-CoV-2, the spike proteins that have two subunits (S1 and S2) mediates virus entry by attaching to the receptor, angiotensin-converting enzyme 2 (ACE2), with the help of hydrolyzing transmembrane protease, serine 2 (TMPRSS2) present on the host cell membrane (Huang et al., 2020). The binding affinity of SARS-CoV-2 RBD to its receptor is 10–20 times higher than that of SARS-CoV (Wrapp et al., 2020, Lan et al., 2020). Hypothetically, pre-existing antibodies formed due to prior SARS-CoV or MERS-CoV and other human coronaviruses infection may recognize the S protein. However, if it does not neutralize the virus, it would result in enhanced illness. In vitro studies conducted to test this hypothesis using sera from immunized mice demonstrated ADE at sub-neutralizing concentrations of antibodies (Jaume et al., 2011a). When SARS-CoV pseudotyped lentiviral particles (PP) were used to infect different cell lines expressing FcγRII but not ACE2 in the presence of sera from mice vaccinated with inactivated SARS-CoV resulted in the increased uptake of PP into FcγRII expressing cell lines (Jaume et al., 2011b). This indicates ADE in SARS-CoV via FcγR mediated mechanism. Also, diluted murine anti-spike antisera enhanced the entry and replication of SARS-CoV in human HL-CZ promyelocytic cell lines that express ACE2 in lower levels and FcγRII in higher levels (Wang et al., 2014). The infectivity assay results showed that ADE in SARS CoV primarily occurs in the presence of diluted antisera, which refers to the sub-neutralizing concentrations of antibodies as the cause of ADE. In contrast, higher concentrations of antiserum neutralized SARS-CoV in mice (Wang et al., 2014). Enhanced infection was also observed in B cell lines in in vitro experiments by SARS-CoV vaccine induced antibodies (Kam et al., 2007). Many other in vitro studies demonstrated that FcR-expressing phagocytes showed enhanced SARS-CoV and MERS-CoV virions uptake when treated in the presence of diluted infected human antisera (Yip et al., 2016, Yip et al., 2014a, Jaume et al., 2011b, Cheung et al., 2005). Vaccination with recombinant full-length spike protein of SARS-CoV provided a protective immunity in macaques. However, an enhanced viral infection of human B lymphocytes was observed by ACE 2-independent and FcγRII-dependent pathways, suggesting ADE in SARS-CoV (Wang et al., 2016a). Neutralizing monoclonal antibody (mAb) against the MERS-CoV RBD enhanced virus uptake into macrophages and other cell lines transfected with FcγRIIa (Wan et al., 2020). For both MERS-CoV and SARS-CoV, a low antibody concentration facilitated ADE, while the higher concentration neutralized the virus (Wan et al., 2020, Jaume et al., 2011c). In SARS-CoV vaccine studies in animal models, vaccinated animals with SARS-CoV demonstrated ERD or increased immunopathology (Deming et al., 2006, Tseng et al., 2012). Ralph Baric et al. observed that Venezuelan equine encephalitis virus replicon particles (VRP) expressed with SARS-CoV N glycoprotein enhanced the infection in mice after homologus and heterologous challenge. Mice demonstrated eosinophilic infiltrates in the lungs starting from day four and persisted till two weeks (Deming et al., 2006). In another study, Robert Couch et al. evaluated four candidate vaccines against SARS-CoV, including recombinant DNA vaccine, virus-like particle (VLP) vaccine, inactivated whole virus vaccine, and spike S protein vaccine. Their findings concluded that Th-2 response of immunopathology was observed in mice given with inactivated vaccine (Tseng et al., 2012). There are only few studies in ADE of MERS, however, in 2018 Prescott et.al reported that antibodies against inactivated MERS virus resulted in hypersensitive lung pathology in rhesus macaques. A vaccine mediated increase in the production of eosinophil granulocytes resulted in interleukin (IL-5 and IL-13) secretion was observed in vaccinated macaques when compared to non-vaccinated (Prescott et al., 2018). Wan et.al., in 2020 demonstrated that the MERS monoclonal antibodies at lower concentrations could exacerbate ADE. The viral entry in to cells expressing FcγR was much higher when compared to DPP4 expressing cells, which are the viral entry receptors (Wan et al., 2020). Interestingly, a recent study based on cellular and structural biology analysis using sera from COVID-19 recovered patients suggest that some RBD-specific antibodies like 7F3 have dual nature: neutralizing pseudovirus and exhibiting ADE in Raji B cells (Wu et al., 2020). This was dependent on the concentration of antibodies and the receptor expression on the cells. It was an IgG-mediated enhancement of infection that is similar to SARS-CoV and MERS-CoV. However, the study reported that optimal antibody concentration blocks the virus entry through the interception of RBD-ACE2 receptor binding. In contrast the sub-neutralizing antibody concentration promoted pseudovirus internalization into the cells expressing FcγR of antibody (Wu et al., 2020). The fact these antibodies to promote the phagocytic uptake of the virus is foreseen; however, the infection of FcγR expressing macrophages is abortive in case of SARS-CoV (Chen et al., 2021) unlike MERS-CoV (Zhou et al., 2014). Similarly, in SARS-CoV-2, ADE as the cause of disease severity has been actively investigated by many researchers. However, reports suggest that in severe COVID-19 patients, antibody titer is high, corresponding to disease severity and mortality (Lau et al., 2021). The reports from the structural analysis suggest that SARS-CoV-2 may escape from the neutralizing antibodies as the number of neutralizing epitopes on the virus is low compared to other RNA viruses (Bachmann et al., 2021). While other RNA viruses including SARS-CoV offer 20 or more repetitive antigenic epitopes that are rigid and induce effective B cell responses, the SARS-CoV-2 offer a smaller number of epitopes and are widely spaced on the S protein and hence the antibodies against them may offer only a short- life. Over activation of complement cascade have been reported in different clinical studies to result in the deposition of immune complexes in the airway tissues, leading to inflammatory lung injury in COVID-19 severe ICU patients. Studies reports that higher levels of complements are associated with disease severity in ICU patients (de Nooijer et al., 2021, Zinellu and Mangoni, 2021, Henry et al., 2021b). This indicates that if ADE exists in SARS-CoV-2 it is more likely to be complement-mediated; however, a definite role of ADE in COVID-19 is not yet reported.
Like a natural infection, vaccination with the live attenuated virus has been shown to induce cross-reactive non-neutralizing antibodies, resulting in increased disease severity (Guzman et al., 2013). ADE after vaccination has been reported in RSV, measles and dengue vaccines (Polack et al., 2002a, Delgado et al., 2009, Borges et al., 2019). In RSV, the incidence of enhanced infection after vaccination came up after the 1960 RSV vaccination, when infants above six months of age were frequently developing RSV infection. Kim et al. in 1969 reported on vaccine-induced ADE in RSV (Kim et al., 1969). For children administrated with formalin-inactivated RSV vaccine (FIRSV vaccine), 80% were hospitalized during RSV season. This study also reported an increased infection rate among children who had maternal antibodies when re-infected with RSV (Kim et al., 1969). The in vitro experiments to evaluate ADE after RSV vaccination demonstrated an enhanced RSV internalization in the presence of RSV-specific monoclonal antibodies into monocytic (U937, THP-1) and macrophage-like cell lines (Osiowy et al., 1994, Krilov et al., 1989b). This indicates that the serum antibodies against RSV are not always protective, however, may enhance the infection when exposed to different strains of RSV. Similarly, severe illness was reported in children infected with the measles virus after being immunized with the measles virus vaccine (Nader et al., 1968, Polack, 2007). This was first reported after introducing formalin-inactivated measles virus (FIMV) vaccines to Europe and the United States in the 1960 s. Upon subsequent exposure to wild-type measles viruses, around 16% of children were severely infected with atypical measles (Carter et al., 1962, Fulginiti et al., 1967, Rauh and Schmidt, 1965). These children developed atypical pneumonia, high fever, and unusual rashes on their skin. As a result, the vaccine was withdrawn in 1967 (Fulginiti et al., 1967, Philadelphia, 2020). A model for the pathogenesis of this atypical infection was proposed by Russell et al. in 2006, which suggests that the H-specific IgG promotes the infection of monocytes and macrophages bearing FcγRII on their surface (Iankov et al., 2006). This relates the atypical measles (ADE) developed in FIMV vaccinated children to the vaccine-generated antibodies that are non-neutralizing. However, there are only a few studies on this, and more investigation is required to explain the pathology and mechanism. The first approved vaccine against Dengue virus, CYT-TDV, is an attenuated tetravalent vaccine composed of yellow fever 17D chimeras and four DENV serotypes. After completion of phase III clinical trials, the vaccine was approved in 2018. However, it was found to cause infection in children continuously over 4–5 years from 18 months after vaccination in 2016 (Halstead, 2018). Studies on the risk-benefit ratio in a seronegative population reported that though the efficacy of the CYT-TDV vaccine is high in terms of antibody titer, however, the serostatus of the patients determines the disease outcome after infection. In case of seronegative vaccine recipients, there was a higher risk of hospitalization compared to seropositive vaccine recipients (Sridhar et al., 2018). A recent study in immune primed rhesus macaques reported that the tetravalent DENV vaccine-induced low titer of neutralizing antibody response and hence the serum antibodies demonstrated higher ADE in BHK cells (McCracken et al., 2021). These evidences indicate that there is a risk of ADE associated with vaccines (especially inactivated vaccines), when patients are re-infected with the same strain or different strains. Hence it is one of the most crucial aspects to be considered while designing vaccines. Table 1 summarises the different studies indicative of ADE due to viral infections or vaccination.
ADE in viral infections has two presumed mechanisms (Fig. 1): (i) increased virus uptake into phagocytic cells via FcγR, resulting in increased infection and replication (extrinsic ADE), or ii) formation of immune complexes (virus-antibody complex), which may lead to complement activation and deposition causing virus-tolerant states leading to increased inflammation (intrinsic ADE) (Lee et al., 2020). This creates airway congestion in respiratory infections and hence leading to ERD (Winarski et al., 2019b). In both ADE mechanisms, FcγR on the surface of the immune cells is the key receptor that promotes enhanced infection. FcγR is the receptor for the Fc portion of immunoglobulin G (IgG) on cells, including macrophages, eosinophils, neutrophils, dendritic cells, B cells, and mast cells. There are different types of FcγR expressed on immune cells that are studied to be associated with ADE in different viral infections and viral vaccines (Pincetic et al., 2014) ( Table 2). ADE mediated by the entry of virus via FcγR uses two different mechanisms by various viruses. The Extrinsic and Intrinsic mechanisms.
The intrinsic ADE seems to have a greater contribution to enhanced illness in flaviviruses infections (DENV, WNV, and Zika virus) as compared to extrinsic ADE. In the intrinsic ADE, augmented virus replication is related to the inhibition of type1 interferon and activation of interleukin-10 biosynthesis, which favors Th2 type immune response. This mechanism is best described using DENV virus. It has also been reported in other viruses, including HIV (Weiss and Scott, 1981, Halstead and O'Rourke, 1977b). DENV has four serotypes (1−4). Infection with one serotype induces protection from future infection with the same serotype. However, subsequent infection with another serotype (heterologous) results in enhanced disease by pre-existing cross-reactive but non-neutralizing antibodies (Dejnirattisai et al., 2010). When non-neutralizing antibodies bind to a virus without preventing or clearing the infection, an extrinsic ADE mechanism may occur. These antibodies bind to the virus's surface glycoproteins, making them more prone to be engulfed by phagocytotic cells (macrophages, monocytes, or DC) through Fc-FcR, specifically FcγR (Narayan and Tripathi, 2020). Once the immunocomplex ligates with the FcγR, the complex enters the endosome of the effector cell. When the affinity for the FcγR to IgG is low, the complex is released inside the cell. This dissociation of immunocomplex from FcγR switches the antiviral innate immune mechanism to an immune-suppressive one. In DENV infection, ADE was more linked to mature DCs than immature DCs, knowing that mature DCs expresses more FcγRIIa but not FcγRIIb (Guilliams et al., 2014). It is also found that FcγRIIa has low affinity to its ligand on IgG (Boonnak et al., 2013, Mohamad Zamberi et al., 2015). The immune suppressive pathway (inhibition of type1 interferon) takes place in two ways (Ubol et al., 2010). The first mechanism is cytokine-mediated. The immunocomplex upregulates the production of TNF, IL-6, and IL-10. High levels of IL-10 activates the cytokine suppressor gene (cytokine signaling genes), further suppress the expression of type 1 interferons, which in turn enables heightened virus production (Taylor et al., 2015). The second mechanism is mediated by negative regulators that help virus replication. The two reported negative regulators are autophagy-related proteins (ATG; 5 and 12) and dihydroxyacetone kinase. When these two are activated, they further deactivate the signal cascade, which supresses type 1 interferon production. Thus interferon-related antiviral responses are affected, and the virus replicates inside the cells resulting in enhanced infection (Ubol et al., 2010).
Intrinsic ADE mechanisms are best studied in respiratory viruses. Respiratory disease enhancement and immunopathology result from increased Fc-mediated antibody-effector functions. The formation of virus-antibody complexes that activate immune cascades and lead to noticeable lung pathology (Winarski et al., 2019b). The activation of immune cells (monocytes, macrophages, dendritic cells, neutrophils, and natural killer cells) by Fc-mediated response by non-neutralizing antibodies can cause dysregulated activation of the immune system (Winarski et al., 2019b, Ye et al., 2017). This ADE mechanism has been extensively investigated in vitro and in vitro via disease manifestations, immunopathology, and presence of inflammatory markers. Non-macrophage-tropic viruses including measles and respiratory syncytial virus (RSV) are clear examples of ADE triggered by increased immune activation. This leads to cytokine and complement pathway activation, which contributes to inflammation, triggering acute respiratory distress syndrome (Polack et al., 2002a, Polack et al., 2003). To further understand the mechanism of non-neutralizing immune complexes to cause enhanced disease, a handful of in vitro studies were performed. A study reported the role of complements, specifically complement 3 (C3), activated by non-neutralizing immune complex (Polack et al., 2002b). The immune-complex formation and interaction with C1q further cleaves C3 into C3a and C3b. C3a is an anaphylatoxin, and C3b is an opsonin. C3b activates down the line complements and gets deposited in the airway tissues (Prohászka et al., 2004). The recruitment of immune cells results in the release of pro-inflammatory cytokines leading to increasing lung pathology (Polack et al., 2003). The released complements and non-neutralized virus-IgG complex gets deposited in the airway tissues causing obstruction leading to acute respiratory distress syndrome in severe infections (Polack et al., 2002a). Many studies reported the role of complement deposition along with immunocomplex associated with severe disease outcome in various respiratory virus infections. A study observed colocalization of IgG and C3 in RSV vaccinated mice's alveolar regions, which demonstrated ERD, unlike control mice. In this study Kim et al. in1976 reported that C3 activates other complements to form complement 5b-9(C5b-9) or activates complements like C3a, C4a, and C5a (anaphylatoxins), which can injure lung tissues. This leads to mucus secretion and congestion of bronchi, and this initiates recruitment of inflammatory cells and disease enhancement (Kim et al., 1976). Another study using six monoclonal antibodies specific to RSV G and F glycoproteins, observed that the immunocomplex initiates a type 2 helper cell (Th2) response that upregulates TNF-α, IL-4, IL-13 and IL-5 expression and down regulates cytotoxic T lymphocytes (Gimenez et al., 1996b). This further leads to dysregulation of the immune system, resulting in non-clearance of virus-infected host cells and enhanced disease. Over activation of the complement followed by the inflammatory lung injury was also observed in SARS and COVID-19 (Wang et al., 2020, Gui, 2020). From the available reports, SARS-CoV-2 is not known to infect macrophages (Narayan and Tripathi, 2020). Thus, the possible mechanism of ADE in SARS-CoV-2 pathology has been explained by the formation of immune complexes that promote excessive immune cascade activation in lung cells (Wu et al., 2020). Clinical evidences reports that excessive complement activation in COVID-19 severe cases and ICU admissions are associated with respiratory failure (Holter et al., 2020a, Chouaki Benmansour et al., 2021, Cugno et al., 2020). Recent studies report that the immune complex may activate C3 that further activates C5a and its convertase. These complements promote the recruitment of macrophages/monocytes and neutrophils. These activated cells secrete proinflammatory cytokines TNF-α and IL-6 that contributes to the cytokine storm (Chouaki Benmansour et al., 2021) resulting in disease severity.
The activation of complements associated with ERD of respiratory viral infections was studied in influenza, RSV, SARS-CoV, and MERS (Garcia et al., 2013). In all the three pathways including classical pathway, lectin pathway and alternative pathway, the central component of the complement cascade is complement 3 (C3). The classical pathway (CP) gets activated when C1q-virus- antibody complex activates C3. C3 gets cleaved to form C3 convertase that further activates other components of the cascade. However, the lectin pathway (LP) gets activated by the activation of Mannose-binding Lectin (MBL) and ficolin complex together named as MBL-associated serine proteases (MBLSPs), when they recognize the carbohydrate patterns on the surface of antigens. MBLSPs cleaves C4 and C2 to form the C4bC2a C3 convertase. In the third pathway, the alternative pathway (AP) the complement cascade is activated by hydrolysis of C3. This pathway also serves as an alternative way to cleave C3 to its products. Both in CP and LP the convertases cleave C3 to C3a and C3b, C3a is an anaphylatoxin and C3b further activates C5 convertases to cleave C5 to C5a and C5b. C5a is an anaphylatoxin and C5b further activates the complements in the cascade. Both these C3a and C5a anaphylatoxins are involved in chemotaxis and cellular effector functions of innate and adaptive immune response (Stoermer and Morrison, 2011). Experiments in mice to study the role of complement 5a (C5a) in ADE revealed that the deficiency of the C5a receptor in mice showed reduced clinical symptoms of influenza infection. They also observed similar effects after blocking the C5aR using specific antibodies (Song et al., 2018). The excess activation of C3a and its role in disease severity were also studied in mice during H5N1 infection (Sun et al., 2013, O'Brien et al., 2011). The study observed a significant reduction in the infiltration of neutrophils and eosinophils in the lungs and reduced viral replication using a C3a receptor (C3aR) antagonist. In SARS CoV infection, a study in mice injected with mouse-adapted SARS-CoV (SARS-CoV MA 15) observed a significant increase in complements, including C3a and C4b, in mice infected with a lethal dose of SARS-CoV (Gralinski et al., 2018). However, silencing of C3 resulted in reduced production of IL-1, IL-6, and TNF-α which are found to be elevated in patients with severe respiratory diseases. The study presents complements as a prime component of disease severity in SARS-CoV infection (Gralinski et al., 2018). A similar investigation was also done in MERS after observing increased levels of C5a in infected patients (Jiang et al., 2018). MERS-CoV infection upregulated the expression of C3a and its receptor (C3aR) in monocytes (THP-1) and macrophages (differentiated THP-1macrohpages) (Jiang et al., 2019). Similarly, in SARS-CoV-2, patients admitted to ICU reported high levels of C5a, C3bc, C3bBbP, C4d, and MAC (Josset et al., 2013). A handful of studies reported the activation of complements via the lectin pathway (Holter et al., 2020b, Gao et al., 2020b, Malaquias et al., 2021). These reports highlight the fact that the ERD in respiratory infections via ADE is complement-mediated though the non-neutralizing antibodies facilitate this phenomenon via FcγR entry in to immune cells.
The complement system consists of approximately 50 proteins that function via three different pathways: i) lectin pathway (LP), ii) classical pathway (CP), and iii) alternative pathway (AP) (Dunkelberger and Song, 2010). The antibody-attached virus activates the complement proteins to form the membrane attack complex (MAC), which destroys the infected cell (Duensing and Watson, 2018). This is known as complement-mediated cytotoxicity. The first compliment in the system to get activated is complement 1q (C1q), which has two subunits C1r and C1s, mainly produced by macrophages, monocytes, and immature dendritic cells. In contrast, the liver produces a majority of the other complement proteins (Merle et al., 2015). C1q is required for the activation of normal IgG responses (Mehlhop and Diamond, 2006). In association with natural IgM or IgG antibodies, C1q activates the classical pathway during primary infection (Stoermer and Morrison, 2011). C1q binding to IgG depends on the clustering of IgG, and this clustering is driven by antigens. Antigens induce the formation of IgG hexamers, promoting multivalent C1q binding on the surface of the antigen-antibody complex, mediated by IgG Fc: Fc interactions, leading to activation of the complement system (Diebolder et al., 2014, Wang et al., 2016b). The primary C1q binding residues are revealed on a platform formed from six Fc portions; as one IgG Fab arm attach to the antigen, the second arm stretches upward towards the C1q stem, resulting in a binding site for hexavalent C1q with high binding avidity (Byrne and Talarico, 2021) ( Fig. 2). C1q binds to virus-antibody complexes, facilitating viral capsule fusion to the cell membrane via the interaction of C1q and its receptor, forming a virus-antibody-C1q complex (von Kietzell et al., 2014). Further, this complex binds to the receptor for C1q on the host cell, triggering the intracellular signaling cascade and enhancing virus-receptor binding. By cleaving C2 and C4 from C1s, C1q can activate the binding of complement C3 and its receptors (Stoermer and Morrison, 2011). Likewise, the virus antibody complex can bind to complement receptors (Dustin, 2016). The mechanism of complement- mediated ADE is also reported in WNV and HIV (Mehlhop et al., 2007). During HIV infection, the C1q protein can bind directly to gp41, a glycoprotein on the outer membrane of HIV. C1q receptors are present on inflammatory monocytes and macrophages, B cells, neutrophils, fibroblasts, smooth muscle cells, endothelial cells, and other cell types. When these antibodies are at sub-neutralizing concentrations for example; the antiviral serum at an early stage may exhibit an increased human monocyte infection. This kind of enhanced infection was reported in HIV by Robinson and his team using MT-2 cells (Robinson et al., 1988b). A similar mechanism was also found in EBOV (von Kietzell et al., 2014). These evidences explain the role of normal levels of C1q in preventing ADE. ( Fig. 3). ADE of homologous and heterologous DENV infection was reported to be C1q-dependent using a mouse model (Yamanaka et al., 2008, Mehlhop et al., 2009b). Mehlhop et al. in 2007 demonstrated that C1q could inhibit ADE both in vitro and in vivo using flavivirus antisera obtained by infecting mice with WNV (Mehlhop et al., 2007). His study reported that C1q can inhibit virus internalization in macrophages in vitro and also in C1q deficient mice when treated using purified C1q. However, it was also shown that ADE was not reduced in the presence of C1q- and C3-depleted serum in vitro, but at high levels of C3 and normal levels of C1q (by reconstituting complement factors to the deficient serum), the rate of ADE was reduced (Yamanaka et al., 2008). All these shreds of evidence indicate the ability of C1q to make changes in the conformation of IgG needed for the fusion to viral E proteins, which promotes the entry of the antibody-bound virus. When C1q is present, the IgG subtype is more determinant of protection (Mehlhop et al., 2007). In the absence of C1q, human IgG1 and IgG3 neutralized WNV better than IgG2 and IgG4. Mehlhop et al. found that the IgG2b class had no impact on the neutralization potential of humanized E16 antibodies in C1q deficient mice. It could be due to the affinity of this IgG subclass for the specific FcγR expressed. The different affinities of different FcR determine the competition between FcR and C1q for IgG subtypes due to overlapping FcR and C1q binding sites (Mehlhop et al., 2007). It was also found that C1q binding to virus surface doesn’t prevent cellular attachment, increasing the possibility that the attachment is mediated by FcR after binding C1q to the antigen-antibody complex (Mehlhop et al., 2009b). His findings concluded that C1q could increase the efficacy of antiviral antibodies by improving the stoichiometric conditions that are required for complete neutralization. In animal models the role of complements was investigated by few studies. These evidences describe the role of C1q or other complements in preventing ADE and moreover its affinity to IgG isotypes, which are vital information in the development of vaccines.
Viral infections are a significant public health concern, as exhibited by the current COVID-19 pandemic. With the absence of effective antivirals and vaccines against many viral infections, it is essential to understand the molecular details of disease prognosis. Regardless of massive vaccination in many countries, COVID-19 cases were in continuous increase. The inactivated SARS-CoV-2 vaccine (BIBP) showed less protective efficacy, reported in Bahrain (Khoury et al., 2021). Later the country offered a booster dose of the mRNA vaccine. From the earlier reports, vaccination with FI-RSV, measles, and other viruses was shown to increase morbidity and mortality after infection due to ADE. Hence, ADE is considered a significant barrier in some viral vaccine development. Understanding the molecular mechanisms of ADE may pave the way towards developing safe vaccine approaches and therapeutic strategies. ADE is a phenomenon widely studied in vitro, however the transferability of its approaches to in vivo and human applications is widely necessary. Animal model studies on RSV and SARS-CoV vaccines indicated the elicitation of Th2- mediated enhanced lung pathology; however, Th1 response is not reported to be associated with such ERD. There is evidence that an adenovirus-based vaccine for MERS S1 fusion protein that could produce Th1 response; preventing vaccine-induced enhanced disease in mice (Hashem et al., 2019). Among all the vaccines against COVID-19 the Adenovirus type 5 expressing spike protein of SARS CoV-2, is reported to be associated with Th1 cytokine, but not with inflammatory cytokine response in immunized BALF (Chung et al., 2022). In the preclinical animal studies, a modified vaccinia virus Ankara (MVA) expressing SARS CoV was reported to demonstrate enhanced hepatitis in ferrets (Weingartl et al., 2004). However, in contest of SARS CoV-2 the possibility of ADE is unclear though some experimental evidences support the same. Animal model studies are essential to understand the risk of enhanced infection at a sub-neutralizing concentration of antibodies against SARS CoV-2. It is very important to understand the need of routine vaccination against SARS CoV-2 considering the emergence of variants as well as the existence of other coronaviruses. This evidence demands the need for more studies using animal models during vaccine development to eliminate the possibility of ADE. The role of complements in homeostasis and innate immunity is pivotal, where antibodies can initiate protective mechanisms against viruses through complex pathways. Studies of ADE in respiratory infections clearly indicates that the ERD due to ADE is complement mediated. These findings indicate that complement activation may contribute to pathogen clearance and inflammation, highlighting complement's dual function. Reports have suggested that patients with pre-existing conditions are more likely to suffer from severe diseases during COVID-19 infection. ADE can be a possible cause of ERD of SARS-CoV-2 in patients with different severities. Vaccines that promote humoral immune responses with predominant protective interactions may be developed considering the circumstances surrounding the interaction between the above-mentioned immune components and the resulting ADE limitation. |
|
PMC9647211 | Shota Ueno,Kenji Kokura,Yasushi Kuromi,Mitsuhiko Osaki,Futoshi Okada,Shinji Kitamura,Tetsuya Ohbayashi | Kidney organoid derived from renal tissue stem cells is a useful tool for histopathological assessment of nephrotoxicity in a cisplatin-induced acute renal tubular injury model | 18-07-2022 | organoids,kidney,cisplatin,immunohistochemistry | Organoids derived from renal tissue stem cells (KS cells) isolated from the S3 segment of adult rat nephrons have previously been developed and evaluated. However, data regarding the histopathological evaluation of these organoids are limited. Therefore, in this study, we performed histopathological examinations of the properties of these organoids and evaluated the nephrotoxicity changes induced by cisplatin treatment. We observe that the tubular structure of the organoids was generally lined by a single layer of cells, in concordance with previous findings. Microvilli were exclusively observed under electron microscopy on the luminal side of this tubular structure. Moreover, the luminal side of the tubular structure was positive for aquaporin-1 (Aqp1), a marker of the proximal renal tubule. Cisplatin treatment induced cell death and degeneration, including cytoplasmic vacuolation, in cells within the tubular structure of the organoids. Cisplatin toxicity is associated with the induction of γ-H2AX (a marker of DNA damage) and the drop of phospho-histone H3 (a marker of cell division) levels. During the nephrotoxicity assessment, the kidney organoids displayed various features similar to those of the natural kidney, suggesting that it is possible to use these organoids in predicting nephrotoxicity. The histological evaluation of the organoids in this study provides insights into the mechanisms underlying nephrotoxicity. | Kidney organoid derived from renal tissue stem cells is a useful tool for histopathological assessment of nephrotoxicity in a cisplatin-induced acute renal tubular injury model
Organoids derived from renal tissue stem cells (KS cells) isolated from the S3 segment of adult rat nephrons have previously been developed and evaluated. However, data regarding the histopathological evaluation of these organoids are limited. Therefore, in this study, we performed histopathological examinations of the properties of these organoids and evaluated the nephrotoxicity changes induced by cisplatin treatment. We observe that the tubular structure of the organoids was generally lined by a single layer of cells, in concordance with previous findings. Microvilli were exclusively observed under electron microscopy on the luminal side of this tubular structure. Moreover, the luminal side of the tubular structure was positive for aquaporin-1 (Aqp1), a marker of the proximal renal tubule. Cisplatin treatment induced cell death and degeneration, including cytoplasmic vacuolation, in cells within the tubular structure of the organoids. Cisplatin toxicity is associated with the induction of γ-H2AX (a marker of DNA damage) and the drop of phospho-histone H3 (a marker of cell division) levels. During the nephrotoxicity assessment, the kidney organoids displayed various features similar to those of the natural kidney, suggesting that it is possible to use these organoids in predicting nephrotoxicity. The histological evaluation of the organoids in this study provides insights into the mechanisms underlying nephrotoxicity.
In vitro assay systems capable of evaluating organ-specific toxicity are warranted. Although previous in vitro studies have made use of conventional culture methods such as monolayers on plates or suspensions in different media, the 3D-structure culture called organoids, which consist of multiple types of cells—including stem cells—capable of imitating organ-specific tissues, have recently been developed, , , . Compared with conventional culture methods, the characteristics of 3D organoids resemble those of the tissue in the living body more closely. Therefore, 3D organoids can be used more accurately to perform toxicity tests, . Though several studies have reported the development of organoids mimicking various tissues, , , , , , , , kidney organoids derived from cell lines (KS cells) isolated from the S3 segment of adult rat renal proximal tubules, by Kitamura et al., are quite unique as they have the nephron-like-structure, . This organoid is characterized by an outward extension of the tubular structure into the extracellular matrix gel, Matrigel. Various analyses of these organoids have been conducted, including fluorescent immunohistochemistry of renal markers in isolated tubular structures, polymerase chain reaction (PCR) analysis of gene expression, and electron microscopy. In addition, the induction of cell death following exposure to cisplatin was confirmed by Kuromi et al. using uptake tests of propidium iodide (PI) (manuscript in preparation). However, to further elucidate the properties of these organoids, additional detailed analyses including morphological evaluations are warranted. As reported by Fujii et al., recently, many scientists are beginning to use histopathological methods to analyse organoids. Detailed morphological observations of cells derived from organoids will provide insights into the molecular mechanisms underlying toxicity, thereby allowing for more accurate prediction of drug toxicity. Therefore, this study aimed at investigating the histopathological characteristics and mRNA expression patterns of this newly developed kidney organoid and analysing the toxic effect of cisplatin-induced renal tubular injury.
The cell line and culture methods used in this study have been previously described by Kitamura et al. In a brief note, KS cells which were transferred from Kitamura et al. were cultured on a type IV collagen plate (Corning Life Sciences, Kennebunk, ME, USA) and maintained in a 1:1 mixture with a conditioned culture supernatant (DMEM [Thermo Fisher Scientific, Waltham, MA, USA] containing 10% fetal calf serum [Thermo Fisher Scientific]) from mouse mesenchymal cells and modified K1 medium at 37°C with 5% CO2/ 100% humidity. The modified K1 medium comprised of a 1:1 mixture of Dulbecco’s Modified Eagle’s Medium (DMEM) and Ham’s F‐12 medium (Thermo Fisher Scientific), supplemented with 10% FCS, 5 µg/mL insulin, 2.75 µg/mL transferrin, 3.35 ng/mL sodium selenious acid (Thermo Fisher Scientific), 50 nM hydrocortisone (Sigma, St. Louis, MO, USA), 25 ng/mL hepatocyte growth factor (Sigma), and 2.5 mM nicotinamide (Sigma). KS cell sheets were incubated with trypsin (Thermo Fisher Scientific) and harvested. Cell clusters were obtained from the KS cells using the “hanging drop” method and these were incubated at 37°C and 5% CO2/ 100% humidity for approximately 6 h. Each cluster contained approximately 1.0 × 105 KS cells. Cell clusters were then transferred into a “half Matrigel” solution situated on the filters of transwell inserts in wells containing the differentiation medium. The “half Matrigel” solution comprised a 1:1 mixture of Matrigel (Corning Life Sciences) and differentiation medium. The differentiation medium contained DMEM/ F‐12 supplemented with 10% FCS, 250 ng/mL glial cell line-derived neurotrophic factor (GDNF) (R&D Systems, Minneapolis, MN, USA), 250 ng/mL basic fibroblast growth factor (bFGF) (R&D Systems), 250 ng/mL epidermal growth factor (EGF) (R&D Systems), 250 ng/mL bone morphogenetic protein-7 (BMP‐7) (R&D Systems), and 250 ng/mL hepatocyte growth factor (HGF) (Sigma). Cell clusters in the “half Matrigel” solution were then cultured for up to 20 days to yield organoids.
Regarding the toxicity analyses, organoids cultured in the “half Matrigel” solution for 14 days were placed in different differentiation media each supplemented with 0, 20, or 30 μM cisplatin (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) for 24, 48, or 144 h and with 10 μM cisplatin for 24 and 144 h. Exposure to cisplatin at 0, 10, and 20 μM for 144 h was performed several times (n=3 or 4). Exposure to all the other conditions was performed once, with n=2. To serve as controls, organoids cultured without cisplatin for the same duration were prepared.
After exposure to cisplatin, the organoids were fixed in 4% paraformaldehyde (FUJIFILM Wako Pure Chemical Corporation) at room temperature for 30 min, washed with phosphate-buffered saline (PBS), and subsequently embedded into a paraffin wax using routine procedures. Paraffin-embedded samples were 4 µm thick and were stained with haematoxylin and eosin (HE) for morphological examination. Also, immunohistochemical analysis and Terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assays were conducted on the samples to further examine their characteristics. The severity of the cytotoxicity was evaluated and scored for the different HE sections of the organoids of each group with different cisplatin exposure doses and durations. The scoring criteria were as follows: − = no dead cells in the tubular structure; + = a few dead cells in the tubular structure (less than 10% of all cells in the tubule); ++ = many dead cells in the tubular structure (approximately 10 to 50% of all cells in the tubule). +++ = over 50% of all cells in the tubular structure were dead.
Immunohistochemical analyses were performed on the sections to confirm the nature of the cells constituting the organoids, , , . Endogenous peroxidase was blocked by treating the sample with 3% hydrogen peroxide in methanol for 12 min. For antigens retrieval, sections were treated with a citrate buffer (pH 6.0) (Agilent, Santa Clara, CA, USA) by heating in a microwave (95°C) for 30 min. After incubation with each primary antibody at 4°C overnight, immunolabeled antigens were visualized using the Simple stain rat Max-PO (Nichirei Bioscience, Tokyo, Japan) and Simple Stain DAB solution (Nichirei Bioscience), and the sections were subsequently counterstained with haematoxylin. Several primary antibodies were used in this study. A polyclonal rabbit anti-aquaporin-1 (Aqp1) antibody (#2219, 1:2,000, Millipore, Bedford, MA, USA) was used as a marker for the renal proximal tubule. A polyclonal rabbit anti-phospho-histone H2A.X (Ser139) antibody (γ-H2AX, #2577, 1:200, Cell Signaling Technology, Danvers, MA, USA) was used for the detection of DNA damage (this antibody was used in the analysis of organoids treated with cisplatin at concentrations of 30 μM for 24 h and 10 μM for 144 h). A monoclonal rabbit anti-Ki-67 antibody (ab16667, 1:100, Abcam, Cambridge, UK) was used for the detection of cell proliferation; this antibody was used in the analysis of organoids with cisplatin at concentration of 10 and 20 μM for 144 h. A polyclonal rabbit anti-phospho-histone H3 (Ser10) antibody (#9701, 1:200, Cell Signaling Technology) was used for the detection of cell division; this antibody was used in the analysis of organoids with cisplatin at a concentration of 30 μM for 144 h. A polyclonal rabbit anti-cleaved caspase 3 (Asp175) antibody (#9661, 1:300, Cell Signaling Technology) was used to detect apoptosis; this antibody was used to analyse organoids treated with cisplatin at a concentration of 20 μM for 144 h. For phospho-Histone H3 and Ki-67, the proportion of positive cells was counted using sections of organoids subjected to 20 μM cisplatin for 144 h (n=4 for both). Three tissue sections of each organoid at different levels were prepared, and five randomly selected fields, including the tubular structure, were imaged using a 20 × objective lens for each section. All cells in the images, including positive cells, were counted, and the proportion of positive cells was determined.
Terminal deoxynucleotidyl TUNEL assay was performed using a commercially available kit from Trevigen (Gaithersburg, MD, USA) following the manufacturer’s protocol. Briefly, deparaffinized and rehydrated tissue sections were washed with PBS, incubated with proteinase K (15 min), washed, and quenched before labeling with biotin-labelled dUTP. The labelling reaction was subsequently stopped by adding a stop buffer, as provided. The tissue sections were then incubated with HRP-conjugated streptavidin for 10 min, washed, and immersed in 3,3′-Diaminobenzidine (DAB) solution for colour development. The sections were counterstained with haematoxylin. The TUNEL assay was performed for the analysis of organoids with cisplatin at a concentration of 10 μM for 144 h.
For electron microscopic examinations, whole organoids were immersed overnight in 0.1 M cacodylate buffer containing 2.5% glutaraldehyde (pH 7.2) and 2% paraformaldehyde. They were then dehydrated using ascending grades of ethanol, embedded in Epon, and finally cut into cubes with 2 mm sides. Ultrathin sections were stained with uranyl acetate and lead citrate and examined using a JEM-1400 transmission electron microscope (Japan Electron Optics Laboratory Co., LTD, Tokyo, Japan) at 80 kV.
Ten organoids cultured for 14 days were prepared and mixed to form a single sample. Two-dimensional (2D) cultured KS cells on a 10-cm plate dish at approximately 70% confluency were also collected as controls. Total RNAs were extracted from whole samples using an RNeasy mini prep kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The total RNA concentration and quality were determined using a spectrophotometer (ND2000; Thermo Fisher Scientific). Complementary DNA (cDNA) was synthesized from the total RNA using a reverse transcription polymerase chain reaction (RT-PCR) kit (TaKaRa, Shiga, Japan), according to the manufacturer’s instructions. Gene expression was analysed via quantitative polymerase chain reaction (qPCR) using primers specific for Aqp1, Mate1, megalin, OCT1 (Slc12a1), and OCT2 (Slc12a2), which are representative proximal tubular markers and transporters. Hprt1 was selected as an internal control. For Aqp1, OCT1, and Hprt1, we obtained previously validated primer sequences from earlier studies, , . For Mate1, megalin, and OCT2, primer sequences were designed using the Basic Local Alignment Search Tool, National Center for Biotechnology Information (BLAST, NCBI) to confirm the specificity of each primer pair. qPCR was performed using THUNDERBIRD® SYBR® qPCR Mix (TOYOBO, Osaka, Japan), according to the manufacturer’s instructions. The reactions were performed in triplicates for each sample using a StepOnePlus Real-Time PCR System (Thermo Fisher Scientific). The cycling conditions were as follows: 95°C for 2 min; 41 cycles of 95 °C for 15 s, 60°C for 1 min, and 95°C for 15 s. Since amplification efficiency was maintained for each primer set, the expression level of each target gene was assessed using the 2−ΔCt equation. The expression levels of each target gene were calculated relative to the Hprt1 levels. Gene expression data were presented as fold-change values in 3D cultures relative to those in 2D cultures. The sequences of the primer sets used in this study were as follows: Aqp1 fwd-5ATT GCA GCG TCA TGT CTG AG, rev-5GAA CTA GGG GCA TCC AAA C; Mate1 fwd-5TCG TGG GCT ACA TTT TCA CCA, rev-5CAC CAC AGG TAC AGG CAA GA; Megalin fwd-5GTT CCA TTG TGG TGC ATC CG, rev-5GGT GAG AAC CAT CGC TCC AT; OCT1 fwd-5TGG CCG TAA GCT CTG TCT CT, rev-5TCA AGG TAT AGC CGG ACA CC; OCT2 fwd-5TGT GCT GTT GCT ACC TGA GA, rev-5CGG TCT GCT TGC TTG ACT TG; Hprt1 fwd-5GCG AAA GTG GAA AAG CCA AGT, rev-5GCC ACA TCA ACA GGA CTC TTG TA.
Significant differences in gene expression, proportion of phospho-histone H3-positive cells, and Ki-67-positive cells were analysed using Student’s t-test. Statistical significance was set at p<0.05.
KS cells were cultured in accordance with the procedures established by Kitamura et al., and the organoids obtained demonstrated a tubular structure radially extending from the center of the cluster, as previously reported. After 20 days of 3D culture, single organoids approached the edge of the 24-well Transwell inserts (Fig. 1A In the images of the tubular structure obtained using electron microscopy (Fig. 1C), short villi were observed on the luminal side (Fig. 1C2, arrows). Contrarily, villi were not observed on the outer side (Fig. 1C3). In general, intercellular adhesion was loose and villi were observed in the gaps.
Immunohistochemistry results showed that Aqp1 (a marker of the renal proximal tubule) was localized in the luminal wall of the tubular structures (Fig. 2A Meanwhile, qPCR gene expression analysis indicated that Aqp1 expression was significantly higher in the whole organoids (3D culture) than in the 2D culture of parental KS cells (Fig. 2B1). In addition, Mate1 (Slc47a1) expression, a transporter involved in the extracellular excretion of cisplatin on the apical side, was significantly more expressed in the 3D organoids compared to the 2D culture (Fig. 2B2). However, OCT1 (Slc22a1) and OCT2 (Slc22a2), two transporters involved in the intracellular uptake of cisplatin from the outer surface, and megalin, a proximal tubule marker, were not expressed by either culture methods (data not shown). Thus, from the viewpoint of morphology, immunohistochemistry, and gene expression, organoids extending tubular structures were similar to renal proximal tubules in the living body, albeit with some differences, such as the lack of OCT1, OCT2, and megalin genes expression.
Typical images demonstrating the tubular structure of the organoids exposed to cisplatin under various conditions are shown in Fig. 3 Histological comparisons were conducted between organoids exposed to different concentrations of cisplatin (0, 20, or 30 μM) and for different exposure periods (24, 48, or 144 h). The results showed that the degree of cytotoxicity increased as the exposure dose and/or exposure period increased. (Fig. 3E, Fig. 3F).
Immunohistochemical analyses using multiple markers were performed to investigate the response of these cells to cisplatin treatment. Immunopositive cells were observed in the groups exposed to cisplatin using antibodies against γ-H2AX, a marker of DNA damage (Fig. 4A
To investigate apoptosis in the organoids, immunohistochemistry of cleaved caspase 3 and a TUNEL assay were performed in organoids exposed to cisplatin (Fig. 5A, Fig. 5B
In this study, we characterized kidney organoids derived from a cell line isolated from the S3 segment of rat renal proximal tubules using histopathological observations and qPCR analysis of renal tubular markers. Our results showed histopathological changes when cisplatin, a typical drug that injures the proximal tubules, was added to the organoids. As previously reported by Kitamura et al., these organoids contained tubular structures, and one layer of cells was consistently observed to line the tubules. Aqp1 expression was confirmed on the inner surface of the tubular structures. Moreover, the gene expression levels of renal tubular markers (such as Aqp1 and Mate1) increased after switching from the 2D to the 3D cultures. Expression of some genes related to renal proximal tubules, such as OCT1 (Slc22a1), OCT2 (Slc22a2), and megalin, was not detected in the organoids. Megalin is known to be expressed at low levels in immature proximal tubules. In the organoids used in this study, immunostaining for Ki-67 (a marker of cell proliferation) and phospho-histone H3 (a marker of cell division) indicated that the cells had some proliferative capacity, suggesting that these cells were probably not fully matured; thus, some of the renal proximal tubular markers were undetected. The polarity of the microvilli in the cells was also displayed using electron microscopy. Although the organoids differed from the proximal tubules in living organisms in some aspects, they had undergone cell differentiation and were useful for toxicity assessment not mediated by OCT or megalin. To the best of our knowledge, there have been no previous histopathological evaluations of cisplatin toxicity using renal organoids derived from rats. Toxicity studies of cisplatin using human renal organoids have been reported by Takasato et al. and Morizane et al., who exposed the organoids to cisplatin for 24 h; however, the present study involved long-term exposures for up to 144 h to doses of cisplatin similar to doses used in their studies. Takasato used cleaved-caspase 3 and Morizane used γ-H2AX and Kim-1 (kidney injury molecule 1: a marker of kidney injury) as indices to detect the toxicity of cisplatin. In the present study, we first compared the morphology of the cells using HE staining and electron microscopy, and then performed immunostaining using not only cleaved caspase-3 and γ-H2AX, but also phospho-histone H3 and Ki-67 (as markers), and the TUNEL method to detect toxicity. This study provides a comprehensive collection of histopathological evidence of cisplatin-induced cytotoxicity. The toxic effects of cisplatin on organoids were both dose- and time-dependent, as confirmed by the scores of the degree of cell injury. As OCT expression was not detected, cisplatin might have permeated the cells by passive diffusion, resulting in cellular injury. DNA damage, cell proliferation with distorted nuclei, and apoptosis were observed under immunohistochemistry. Immunopositive staining for the DNA damage marker γ-H2AX was observed consistently following cisplatin treatment. Some of the immuno-positive stainings of γ-H2AX were punctate, but the other stainings were often widely distributed throughout the nucleus (were not punctate). Various distribution patterns of γ-H2AX have been reported by Bonner et al. The typical distribution of γ-H2AX in DNA damage is punctate. However, pan-nuclear positivity for γ-H2AX was often observed in the apoptotic cells. These findings suggest that γ-H2AX positivity in cisplatin-exposed organoids may be the result of both DNA damage and apoptosis. Atypical cells with significantly distorted nuclei positive for Ki-67 (a marker of cell proliferation) and phospho-histone H3 (a marker of cell division) were observed in the cisplatin-exposed organoids. Bunel et al. reported that G2/M phase arrest occurred as a result of DNA damage caused by cisplatin, while the proportion of G2/M phase cells increased. This finding suggests that exposure to cisplatin causes G2/M phase arrest in these organoids, and cells with atypical nuclei can be recognized prominently. In the semi-quantitative evaluation of phospho-histone H3, the number of immunopositive cells for phospho-histone H3 was lower in the cisplatin-exposed organoids, suggesting that cisplatin suppressed cell division. However, semi-quantitative analysis of the Ki-67-positive cells showed no significant difference between organoids treated with cisplatin and controls, although there was a decreasing trend of the rate of positive cells in the cisplatin-treated organoids. Since these two markers are found in proliferating cells, the evidence is not strong enough to conclude that cisplatin reduces the rate of cell proliferation. TUNEL test and immunohistochemical analysis of cleaved caspase 3 (markers of apoptosis) revealed that immunopositive sites were mainly located at the interior of the tubular structure of cisplatin-exposed organoids. This positive immunostaining was extensively associated with the remnants of dead cells trapped inside the tubule, suggesting that these cells underwent apoptosis. According to Anada et al., cell death at the center of organoids occurs due to an insufficient oxygen supply. In concordance with this finding, significant cell death was observed in the central segment of the organoids in the absence of cisplatin (Fig. 1B1, arrow). It is possible that oxygen and nutrients may not be sufficiently accessible to the inside of the single-cell lining. Similarly, oxygen and medium ingredients may be lacking in the cells inside the tubular structure, compromising the viability of these cells. Consequently, the susceptibility of cells inside the tubule to cisplatin may be higher than that of the healthy surface cells. In this study, we detected the toxicity effects (such as abnormal cell morphology, DNA damage, apoptosis, and proliferative cell abnormalities) of cisplatin at the cellular level by using conventional histopathology. These toxicological findings suggest that this culture system could serve as a model to evaluate the toxicity of cisplatin on the renal proximal tubules of rats. However, many challenges remain, ranging from the quantitative evaluation of the toxicity to the extrapolation of the results to living organs. Thus, the utility of these organoids as toxicity prediction models warrants further investigation. Toxic substances other than cisplatin should be evaluated using this system. Worth noting, this kidney organoid does not fully mimic the proximal tubule, as indicated by the lack of expression of some renal markers. Therefore, we must first confirm whether the test substance passes through the transporters or is expressed in the organoids. Nonetheless, important toxicological findings can be obtained by using this in vitro system. Our results show that this kidney organoid is a useful system for the assessment of nephrotoxicity, and its histological evaluation will help to fully elucidate the mechanisms underlying nephrotoxicity. This in vitro system can also be incorporated as a screening tool during drug development. This study contributes to the development of alternative methods to animal experimentations.
The authors declare that they have no competing interests. |
|
PMC9647215 | Yuko Yamaguchi,Tsubasa Saito,Mizuho Takagi,Tomomi Nakazawa,Kazutoshi Tamura | Changes in 5-Fluorouracil-induced external granular cell damage during the time-course of the developing cerebellum of infant rats | 30-05-2022 | 5-Fluorouracil,external granular cell,apoptosis,cell cycle arrest,immunohistochemistry,reverse transcription polymerase chain reaction (RT-PCR) | 5-Fluorouracil (5-FU) is widely used as a chemotherapeutic agent that blocks DNA synthesis and replication by inhibiting thymidylate synthetase. This study aimed to elucidate 5-FU-induced changes in the external granular cells (EGCs) in the cerebellum of infant rats and the possible underlying mechanism. Six-day-old infant rats were injected subcutaneously with 40 mg/kg of 5-FU, and their cerebellums were examined at 6, 9, 12, and 24 h after treatment (HAT), and 2, 4, and 10 d after treatment (DAT). The width of the external granular layer (EGL) decreased from 24 HAT to 4 DAT in the 5-FU group compared to that in the control group. However, the width in the 5-FU group was comparable to that of the control group at 10 DAT. The number of apoptotic cells, cleaved caspase 3-labeling index (LI%), p21cip1-LI%, and expression levels of p53, p21cip1, and Fas mRNAs increased at 24 HAT. However, no changes were detected in the expression levels of Puma and Bax mRNAs at any time point. BrdU-LI% increased at 6 and 12 HAT but decreased at 24 HAT. The phospho-histone H3-LI% decreased from 6 HAT to 2 DAT. The width of the molecular layer decreased compared to that of the control group at 10 DAT. No differences were observed in Purkinje cell development. These results indicate that 5-FU inhibited cell proliferation by inducing apoptosis of EGCs via activation of Fas and caspase-3 without the involvement of the mitochondrial pathway and induced p53-dependent G1-S and G2-M phase arrest. | Changes in 5-Fluorouracil-induced external granular cell damage during the time-course of the developing cerebellum of infant rats
5-Fluorouracil (5-FU) is widely used as a chemotherapeutic agent that blocks DNA synthesis and replication by inhibiting thymidylate synthetase. This study aimed to elucidate 5-FU-induced changes in the external granular cells (EGCs) in the cerebellum of infant rats and the possible underlying mechanism. Six-day-old infant rats were injected subcutaneously with 40 mg/kg of 5-FU, and their cerebellums were examined at 6, 9, 12, and 24 h after treatment (HAT), and 2, 4, and 10 d after treatment (DAT). The width of the external granular layer (EGL) decreased from 24 HAT to 4 DAT in the 5-FU group compared to that in the control group. However, the width in the 5-FU group was comparable to that of the control group at 10 DAT. The number of apoptotic cells, cleaved caspase 3-labeling index (LI%), p21cip1-LI%, and expression levels of p53, p21cip1, and Fas mRNAs increased at 24 HAT. However, no changes were detected in the expression levels of Puma and Bax mRNAs at any time point. BrdU-LI% increased at 6 and 12 HAT but decreased at 24 HAT. The phospho-histone H3-LI% decreased from 6 HAT to 2 DAT. The width of the molecular layer decreased compared to that of the control group at 10 DAT. No differences were observed in Purkinje cell development. These results indicate that 5-FU inhibited cell proliferation by inducing apoptosis of EGCs via activation of Fas and caspase-3 without the involvement of the mitochondrial pathway and induced p53-dependent G1-S and G2-M phase arrest.
As a widely used chemotherapeutic agent, 5-Fluorouracil (5-FU) blocks DNA synthesis and replication via inhibition of thymidylate synthetase (TS) and incorporation of its metabolites into RNA and DNA. 5-FU is absorbed rapidly into the maternal circulation, and its metabolites are directly incorporated into embryonic nucleic acid. Since 5-FU readily crosses the blood-brain barrier (BBB), it induces teratogenic effects and subsequent developmental anomalies in the brain of rodents and humans, , , , , . Moreover, there is some evidence showing that 5-FU can cross the BBB by simple diffusion and exert neurotoxic effects, thereby leading to nystagmus, ataxia, dysarthria, and epilepsy in humans, . Although several DNA-damaging agents have demonstrated toxic effects on the developing brains of fetuses and newborns of rats and mice, 5-FU has not been adequately investigated for its toxic effects on the developing central nervous system (CNS). Few detailed reports have been published examining the effects and mechanism of DNA-damaging agents, , , , , , including 5-FU, in the developing cerebellum; however, the timings of administration of such agents were disparate in the aforementioned studies. In our previous study, we elucidated that p53-mediated apoptosis and growth inhibition in neural progenitor cells in the telencephalic wall occur in fetal rats following the administration of 5-FU in pregnant rats on gestational day 13. As the next step, the present study was carried out to clarify the effects and mechanisms of 5-FU on neural progenitor cells during the development of the external granule cells (EGCs) of the cerebellum as well as the cerebrum. Cerebellar granule cells have been widely used as in vitro models to elucidate the mechanisms of action of various therapeutic agents, , , , . Since granule cells continue to develop after birth, , it is inferred that neonatal granule cells are highly sensitive to 5-FU. The cerebellum differs from the cerebrum in its developmental pattern, including neuronal migration and arrangement. Immature neural cells in the cerebellum develop in at least two different germinal zones; Purkinje cells and neurons originate from the ventricular zone and migrate toward the surface just beneath the molecular layer, and granular cells originate in the rhombic lip and migrate across the surface of the anlagen to the external granular layer (EGL) located just below the pia mater, . These cells migrate towards the deep cerebellar cortex, and finally, the cortical layer structure of the cerebellum is formed, . Considering the characteristics of neural cells in the developing cerebellum, the present study focused on the effects of 5-FU on the EGL of the developing cerebellum, and the primary fissure of the vermis was selected as the observation area (Fig. 1
Fifteen pregnant Sprague-Dawley (Crl:CD) rats were purchased from Charles River Japan Inc. (Atsugi Breeding Center, Kanagawa, Japan) on day 13 of gestation. The date of the birth of progenies was defined as postnatal day (PND) 0. All newborn rats were separated from their dams on PND 4 and those in good health were pooled and assigned to the study. Thirteen foster mothers were selected based on their health and nursing conditions, and each mother was allowed to suckle ten infant rats. Ten infant rats from each foster mother were randomly assigned to the control group (n=5) and the 5-FU group (n=5). All animals were housed in family units in plastic Econ cages (W 340 mm × D 450 mm × H 185 mm) with bedding (ALPHA-dri, Shepherd Specialty Papers, Inc. Richland, MI, USA) and maintained in a barrier-sustained animal room controlled at 23 ± 3°C and 50 ± 20% relative humidity, with 10 to 15 times per hour ventilation and a 12 h/12 h light/dark cycle. All dams were allowed free access to sterilized basal diet (CRF-1, Oriental Yeast Co., Ltd., Tokyo, Japan) and tap water. The experiments were carried out in accordance with the Guide for Animal Experimentation of the BoZo Research Center, Inc.
5-FU (FUJIFILM Wako Pure Chemicals, Osaka, Japan) and 5-Bromo-2’-deoxyuridine (BrdU) (Sigma-Aldrich Japan, Tokyo, Japan) were dissolved in saline solution at a dose of 40 mg/kg (10 mL/kg body weight) and 100 mg/kg (10 mL/kg body weight), respectively.
On PND 6, infant rats in the treatment group were injected subcutaneously with 5-FU at a dose of 40 mg/kg, and those in the control group were injected with 10 mL/kg of saline solution. The number of infant rats at each time point was each five in the control and 5-FU groups. However, after ten days of treatment, there were five rats in each group (1 dam). The dose of 5-FU was decided based on the results of our preliminary study. In the preliminary study, six-day-old rats were injected with 5-FU at doses of 30, 40, and 50 mg/kg, and the degree of apoptosis was observed in all treatment groups that were administered various doses at 9 and 24 h after treatment (HAT). The results showed that the degree of apoptosis was weak in the 30 mg/kg group. However, it was extremely severe at 9 HAT in the 50 mg/kg group. As these doses were not suitable for analyzing the time course, a dose of 40 mg/kg was selected for the present study. Infant rats in the control and 5-FU groups were euthanized at 3, 6, 12, and 24 HAT and 2, 4, and 10 d after treatment (DAT) by performing exsanguination from the abdominal aorta under isoflurane anesthesia. Half of the infant rats were subcutaneously injected with BrdU at the same time as saline or 5-FU treatment to observe the migration of EGCs, and the remaining infant rats were injected subcutaneously with BrdU 30 min before euthanasia to detect S-phase cells. At each time point, all infant rats were dissected, and their brains were weighed. The left hemisphere of the brain was fixed in 10% buffered formalin, embedded in paraffin wax, and sectioned for histopathological and immunohistochemical examination. The right hemisphere of the cerebellum was frozen in liquid nitrogen and stored at −80°C for real-time RT-PCR analysis.
For real-time RT-PCR analysis, the right hemisphere of the cerebellum was acquired from five infant rats at each time point (3, 6, 12, and 24 HAT and 2 and 4 DAT) both in the control group and 5-FU group and stored at −80 °C until RNA extraction. RNA was extracted from the samples of three infant rats in both groups at each time point. Total RNA was extracted using the RNeasy Mini Kit (Qiagen, Tokyo, Japan). First-strand cDNA was then synthesized from total RNA by reverse transcription using Taqman® Reverse Transcription Reagent (Applied Biosystems, Carlsbad, CA, USA). For real-time RT-PCR, the reaction mixture contained Power SYBR® Green PCR Master Mix (Applied Biosystems) and sense and antisense primers. The cDNA samples were preheated at 95 °C for 10 min and were subjected to 40 cycles of amplification (denaturation at 95 °C for 15 s, annealing, and extension at 60 °C for 60 s) using the StepOnePlusTM Real-Time PCR System (Applied Biosystems). PCR was performed using oligonucleotide primer sets corresponding to the cDNA sequences (p53, p21cip1, Puma, Bax, Fas), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was as an internal standard (Table 1 The expression levels of mRNAs corresponding to p53, p21cip1, Puma, Bax, and Fas were normalized to those of the internal standard Gadph. The fold-changes relative to the control group values at each point are represented as the mean ± standard deviation (SD) of the values corresponding to three infants.
The left hemisphere of the cerebellum was trimmed longitudinally in the central vermis. The tissues were processed into paraffin-embedded blocks, sectioned at 2 µm, and stained with hematoxylin and eosin (HE).
Paraffin sections of the left hemisphere of the cerebellum derived from all infant rats per group at each time point were used for immunohistochemistry. Paraffin sections were deparaffinized, treated with 0.3% H2O2 in methanol at room temperature for 10 min, and then incubated with protein blocking buffer (Abcam, Boston, MA, USA) at room temperature for 10 min to block nonspecific reactions. Antigen retrieval was performed by heating the sections using a microwave at 95°C for 10 min in 10mM citrate buffer with a pH of 6.0. Immunohistochemistry analyses were performed for examining the expression of cleaved caspase-3 (a marker for apoptotic cells), p53, p21cip1, phospho-histone H3 (a marker for M phase), BrdU (a marker for S phase), and calbindin, as described below. The sections were incubated with the following primary antibodies overnight at 4 °C: rabbit anti-cleaved caspase-3 polyclonal antibody (1:200, Cell Signaling Technology, Tokyo, Japan), rabbit anti-p21cip1 monoclonal antibody (1:100, Dako Japan, Tokyo, Japan), mouse anti-p53 polyclonal antibody (1:1,000, Santa Cruz Biotechnology, Dallas, TX, USA), rabbit anti-phospho-histone H3 polyclonal antibody (1:150, Cell Signaling Technology, Beverly, MA, USA), mouse anti-BrdU monoclonal antibody (1:200, Dako Japan), and rabbit anti-calbindin polyclonal antibody (1:800; Dako Japan). After washing, the sections were treated using the Envision+kit (Dako Japan) at room temperature for 60 min. Only for the anti-BrdU antibody, the sections were incubated with 2N HCl at room temperature for 30 min and with 0.05% protease (Protease type XXIV, Sigma-Aldrich Japan) at room temperature for 5 min before allowing to react with the primary antibody. Positive cells were visualized by performing a peroxidase-diaminobenzidine (DAB; Dojindo Laboratories, Kumamoto, Japan) reaction and counterstaining with hematoxylin.
In HE-stained specimens, the number of pyknotic cells in the EGL at the primary fissure was counted with an upper limit of 300 cells in a field of view of 400×. The apoptosis index was calculated as the percentage of pyknotic cells among the total number of counted EGCs. The widths of the EGL and the molecular layer at the primary fissure were measured using the analytical model FlvFs-LS (Olympus, Tokyo, Japan). In immunohistochemical specimens prepared for analyzing cleaved caspase-3, p53, p21cip1, phospho-histone H3 in four infant rats, and BrdU in three infant rats at each time point, morphometric analyses were performed at the same site at all time points except 10 DAT. For the positive rates of each antibody without calbindin, counting and analysis were performed in the same manner as that of the apoptosis index. Calbindin-stained specimens were used to detect abnormalities in the development of Purkinje cells and dendrites in the molecular layer.
The brain weights and labeling indices (LIs%) of apoptotic EGCs and positive expression rates of cleaved caspase-3, p53, p21cip1, phospho-histone H3, and BrdU in EGCs were expressed as the mean ± SD. The comparisons of the brain weight, widths of the EGL and molecular layer, and all labeling indices between the 5-FU and control groups at each time point were performed using the F-test, followed by a two-tailed Student’s t-test and/or Welch’s t-test. For all comparisons, p-values less than 5% (p<0.05) and 1% (p<0.01) were considered statistically significant.
No deaths occurred in dams or infant rats in any group at any time point and no noticeable clinical sign appeared. No abnormal macroscopic findings were observed in the 5-FU group at any time point during necropsy. However, the brain weights reduced significantly in the 5-FU group at 24 HAT compared to those in the control group (Fig. 2
The expression levels of mRNAs corresponding to p53, p21cip1, Fas, Puma, and Bax were measured by real-time RT-PCR. Among them, the expression level of p53 mRNA significantly increased at 24 and 4 DAT and that of p21cip1 and Fas significantly increased at 24 HAT compared to those observed in the control group (Fig. 3
During the normal developmental process (Figs. 4A and 5A The apoptosis index (pyknotic EGCs) (Fig. 5B Although phospho-histone H3-positive mitotic EGCs were detected throughout the experimental period in the control group, the number of phospho-histone H3-positive EGCs in the 5-FU group decreased from 3 HAT, was less than 0.3% from 12 HAT to 2 DAT, and returned to the control levels observed at 4 DAT (Figs. 5D and 6E, 6F). Few or no p21cip1-positive EGCs were observed in the control group throughout the experimental period (Fig. 5E). In the 5-FU group, p21cip1-LI% significantly increased at 24 HAT (Figs. 5E and 6G, 6H) when the cleaved caspase-3-LI% reached its maximal level (Fig. 5C). BrdU-positive EGCs were observed at all time points in the control group (Fig. 5F). In the 5-FU group, BrdU-LI% increased at 6 and 12 HAT but decreased at 24 HAT and then returned to the control level 2 DAT (Figs. 5F and 6I, 6J). Chronological analysis of BrdU immunohistochemistry in the control group revealed that BrdU-positive EGCs were observed on the pia mater side of the center area of EGL (Fig. 7A Throughout the experimental period, no apparent differences were observed in the cellularity or morphology of Purkinje cells in the 5-FU group compared to that in the control group (Figs. 4and8 The width of the molecular layer increased in the control group from 3 to 10 DAT along with normal development. In the 5-FU group, the molecular layer was thinner than that in the control group at 10 DAT (Figs. 4and9
In the present study, the number of pyknotic EGCs began to increase at 6 HAT and peaked at 24 HAT in the 5-FU group. Most pyknotic EGCs are immunohistochemically positive for cleaved caspase-3. The changes observed in the cleaved caspase-3-LI%, a marker of apoptosis, in the time course corresponded well to those of the aforementioned pyknotic EGCs. Therefore, the presence of pyknotic EGCs observed in this study can likely be attributed to apoptosis. It is widely known that p53 plays a crucial role in apoptosis in response to DNA damage, , . Three processes have been postulated for apoptosis: induction, determination, and execution. Death ligand-mediated and mitochondria-mediated pathways play major roles in the process of apoptotic determination, , , . The BH3-only subfamily is responsible for sensing a wide range of apoptotic stimuli and transmitting this signal to other Bcl-2 proteins to initiate apoptosis. The leakage of cytochrome c from the mitochondria to the cytoplasm is determined by the balance between the expression of Bax and BH3-only proteins, such as Bid and Puma, which increase permeability, and Bcl-2 and Mcl-1 which inhibit apoptosis, . Cytochrome c from the mitochondria binds to Apaf-1 and activates caspase-9, and caspase-9 activates caspase-3 and caspase-7. In contrast, the Fas/Fas ligand activates the receptor complex (death-inducing signaling complex) and activates caspase 8, which activates the lower caspases 3 and 7 involved in the execution. p53 may be responsible for 5-FU-induced apoptosis in human cancer cells, , . However, since the expression of p53 varies in cancer cells, it is unclear whether its expression and mechanism are the same as those observed in normal EGCs. Esperanza et al. reported that 5-FU-induced apoptosis of the cells of the normal thymus in mice is associated with the co-expression of Fas, Bax, and caspase-3. Apoptosis of granule cells in the cerebellum after intraperitoneal administration of 5-FU to seven-day-old rats has been attributed to the activation of the apoptotic pathway of caspase-3. In the present study, real-time RT-PCR analysis revealed that the expression levels of p53 at 24 HAT and 4 DAT and Fas at 24 HAT increased significantly in the 5-FU group, whereas no difference in the expression levels of Puma and Bax was detected between the control and 5-FU groups at any time point. These results are almost the same as those presented in the aforementioned reports, indicating that apoptosis is not mediated by the mitochondrial pathway and that Fas-activated caspase-8 activates caspase-3, resulting in apoptosis without involving the mitochondrial pathway. The reason for high levels of p53 expression observed at 4 DAT in the 5-FU group and the role of p53 in the apoptosis of EGCs under present experimental conditions have not been clarified. In the 5-FU group, phospho-histone H3-LI% was significantly decreased at 6 and 12 HAT and 2 DAT, and BrdU-LI% was significantly decreased at 24 HAT. In contrast, the cleaved caspase-3-LI%, indicating the apoptosis index, increased from 6 HAT, peaked at 24 HAT, and returned to control levels at 2 DAT. These results indicate that 5-FU not only induces apoptosis but also suppresses cell proliferative activity, resulting in a reduction in the width of the external granular layer and brain weight. It is well known that the expression of p53 is induced by DNA damage and various stresses. It functions as a transcriptional activator in the nucleus and plays a crucial role in cell cycle arrest, induction of apoptosis, and DNA repair. Furthermore, p53 is activated in cells with DNA damage, resulting in cell cycle arrest in G1 and G2/M phases. During this period, it inhibits DNA mutation by facilitating DNA repair. P21 is a downstream target of p53 and a potent cyclin-dependent kinase inhibitor that functions as a regulator of cell cycle progression in the G1-S and M phases, , . BrdU-LI% was increased at 6 and 12 HAT, but p21cip1-LI% significantly increased at 24 HAT when BrdU-LI% decreased. In addition, real-time RT-PCR analysis revealed that the expression levels of mRNAs corresponding to p53 and p21cip1 were significantly increased at 24 HAT. Shuey et al., reported that 5-FU induced an increase in the number of S-phase cells at 8 HAT and a remarkable decrease at 24 HAT in the liver of fetal rats, and the peak in the inhibition of TS activity observed in the liver at 24 HAT may be attributed to these cell cycle effects. In the present study, an increase of BrdU-positive S phase cells at the early time points, as described in the aforementioned reference, is likely to reflect the accumulation of S phase cells (late G1 phase cells) until 24 HAT, when a peak of the inhibition of TS activity was observed. These results suggest that 5-FU induced p53-dependent accumulation of cells in the S phase and arrest of EGCs in the G1-S and G2-M phase, resulting in a reduction in the number of mitotic and S phase cells observed in the present study. The similarities between changes observed in the 5-FU group in the present study and those observed in our previous study investigating the effects of 5-FU in neural progenitor cells in the fetal telencephalic wall are outlined as follows: (1) mitosis was remarkably reduced, and G2-M phase arrest occurred from the early stages of treatment followed by apoptosis and G1-S phase arrest; (2) cell proliferative activity was reduced due to a decrease in the number of S phase cells. Some differences in the changes observed in the 5-FU group in these two studies were also noted. In a previous study, the apoptosis index of the neural progenitor cells of the telencephalic wall that were treated with 5-FU showed high values from 9 HAT to 24 HAT and then gradually decreased, strongly suggesting that apoptosis was mediated by p53 because the p53-LI% was significantly higher in the 5-FU group prior to apoptosis. In the present study, the apoptosis index in EGCs treated with 5-FU increased gradually from 6 HAT, peaked at 24 HAT, and then decreased sharply at 2 DAT. However, p53 levels were not significantly higher in the 5-FU group in the present study prior to apoptosis, and the association between apoptosis and p53 expression was not clear. In addition, here, S-phase accumulation was observed at 6 and 12 HAT in the 5-FU group. These differences indicate that 5-FU-induced apoptosis in EGCs was less severe than that observed in neural progenitor cells of the telencephalic wall and that cellular arrest was more apparent in EGCs. Thus, such differences in the degree and duration of apoptosis and the association between apoptosis and p53 are likely to be attributed to the differences in fetal and neonatal exposure to treatment that was administered via different routes and at varying doses in these two studies. No difference was observed in the development of Purkinje cells between the 5-FU and control groups at any time point. Bejar et al. reported that changes in Bergmann glial cells induced by the mitochondrial inhibitor, methylazoxymethanol, were drastic in mice when they were treated at an age of 0 days, whereas the structure was maintained in mice treated at an age of 5 days. They speculated that this difference was due to the date of the treatment. Purkinje cells undergo complete terminal differentiation from day 10 to 13 of gestation and migrate to the cerebellar cortex (parenchyma) by day 16 or 17 of gestation. After birth, cell bodies develop until nine days and dendrites develop rapidly after nine days. Therefore, at the time of administration in this study (six-day-old rats), it is conceivable that Purkinje cells were not affected by 5-FU treatment because they were already in the stage of cell body development and had no proliferative activity. A few BrdU (simultaneous administration)-positive cells in the inner granular layer observed 4 DAT are likely to reflect the migration of surviving EGCs into the inner granular layer. Therefore, under the experimental conditions of our study, it is likely that 5-FU did not affect the migration of external granule cells. However, because the width of the molecular layer in the 5-FU group was thinner at 10 DAT, the effect of 5-FU in the molecular layer may become clearer as time progresses. At 10 DAT, the width and cell density of the external and internal granular layers in some 5-FU-treated rats were similar to those of the control group. This finding suggests that remarkable regeneration of granule cells can be observed between 4 and 10 DAT (10–16 days of age) and is a crucial event. In contrast, the reason for the increased expression of p53 mRNA observed at 4 DAT is unclear; however, this fluctuation may reflect events observed in other areas of the cerebellum. In conclusion, this study elucidated the changes observed in the time-course of EGCs in the vermis of the cerebellum after a single subcutaneous administration of 5-FU to six-day-old infant rats. 5-FU induced the apoptosis of EGCs by activating the Fas and caspase-3 pathways without involving the mitochondrial pathway and led to the p53-dependent accumulation of cells in the S phase, thereby leading to G1-S and G2-M phase arrest. This finding indicates that 5-FU inhibited the proliferative activity of EGCs. EGCs recovered remarkably from 4 to 10 DAT. Furthermore, 5-FU, when administered at an age of six days, may not affect the development of Purkinje cells or migration of EGCs. Although it is well known that 5-FU induces apoptosis in normal and cancer cells, the mechanisms are not the same and are poorly understood, , , , . This study provides useful information for elucidating the mechanisms underlying CNS malformations, adult neurotoxicity, and p53-independent apoptosis in human cancer cells induced by 5-FU.
The authors declare no conflicts of interest directly relevant to the content of this article. |
|
PMC9647221 | Saeed Ur Rahman,Woo-Jin Kim,Shin Hye Chung,Kyung Mi Woo | Nanofibrous topography-driven altered responsiveness to Wnt5a mediates the three-dimensional polarization of odontoblasts | 02-11-2022 | Nanofibrous topography,Cytoskeleton,Cdc42,Wnt5a,Odontoblast | Cell differentiation with the proper three-dimensional (3-D) structure is critical for cells to carry out their cellular functions in tissues. Odontoblasts derived from neural crest cells are elongated and polarized with the cell process, which is decisive for one directional tubular dentin formation. Here, we report that the fibrous topography of scaffolds directs odontoblast-lineage cells to differentiate to have the 3-D structure of odontoblasts through an altered responsiveness to Wnt family member 5A (Wnt5a). In a pulp-exposure animal model, the scaffolds with the nanofibrous topography supported the regeneration of tubular dentin with odontoblast processes. In cultures of pre-odontoblast cells, the nanofibrous topography heightened the cells on the z-axis. The cells on nanofibrous substrate (FIBER) formed stress fiber cytoskeletons on a conventional tissue culture plate (TCP). Differential activation of Cell division control protein 42 (Cdc42) on FIBER and Ras homolog family member A (RhoA) on TCP led to these differences. The signal from Wnt5a-Cdc42 in the cells on FIBER mediated the phosphorylation of JNK and the polarity growth signaling. Taken together, the nanofibrous topography of the scaffolds led to the 3-D structural differentiation of odontoblasts in vitro and in vivo, implying its application for dentin regeneration. Furthermore, the results on the altered activation of Cdc42 by Wnt5a on FIBER provide evidence that the topography of the scaffolds can cause a distinctive cell responsiveness to their micro-environments. | Nanofibrous topography-driven altered responsiveness to Wnt5a mediates the three-dimensional polarization of odontoblasts
Cell differentiation with the proper three-dimensional (3-D) structure is critical for cells to carry out their cellular functions in tissues. Odontoblasts derived from neural crest cells are elongated and polarized with the cell process, which is decisive for one directional tubular dentin formation. Here, we report that the fibrous topography of scaffolds directs odontoblast-lineage cells to differentiate to have the 3-D structure of odontoblasts through an altered responsiveness to Wnt family member 5A (Wnt5a). In a pulp-exposure animal model, the scaffolds with the nanofibrous topography supported the regeneration of tubular dentin with odontoblast processes. In cultures of pre-odontoblast cells, the nanofibrous topography heightened the cells on the z-axis. The cells on nanofibrous substrate (FIBER) formed stress fiber cytoskeletons on a conventional tissue culture plate (TCP). Differential activation of Cell division control protein 42 (Cdc42) on FIBER and Ras homolog family member A (RhoA) on TCP led to these differences. The signal from Wnt5a-Cdc42 in the cells on FIBER mediated the phosphorylation of JNK and the polarity growth signaling. Taken together, the nanofibrous topography of the scaffolds led to the 3-D structural differentiation of odontoblasts in vitro and in vivo, implying its application for dentin regeneration. Furthermore, the results on the altered activation of Cdc42 by Wnt5a on FIBER provide evidence that the topography of the scaffolds can cause a distinctive cell responsiveness to their micro-environments.
Nano-topographic approaches have been developed to simulate physical and biochemical micro-environments that mimic the natural cellular milieu. Physical topographies are associated biologically with cell adhesion. The rationale to these links is the nano- or micro-scale physical topographies of the native ECM. Therefore, structures of engineered scaffolds on relevant scales would be a promising strategy to mimic the natural extracellular matrix (ECM) [1]. A number of approaches, including electrospinning, lithography, embossing, and micromachining, can achieve an engineered ECM with defined topographical features [2]. These approaches can produce micro- or nano-scaled features. The electrospinning technique has gained in popularity due to its ability to produce nanoscale topographies that can influence the cell adhesion, survival, and reorganization outcomes [3,4]. The nanofibrous matrix utilized in tissue engineering has several desirable properties including protein absorption, a high surface area, binding sites for cellular interactions, customized contiguity, the activation of specific intracellular signaling, and different gene expressions. Essentially, these engineered ECM materials can induce cell adhesion, proliferation and differentiation by altering their surface nanotopography. It has been shown that cells respond well to nanotopographic features on synthetic matrix surfaces that induce changes in the cell adhesion and cell differentiation [[5], [6], [7], [8], [9]]. The fibrous structure of electrospun nanofibers can enhance the differentiation of odontoblasts. Previously, it was shown that an electrospun nanofibrous matrix induces the odontogenic differentiation of dental pulp stem cells, specifically dspp expression [6]. Meanwhile, cell differentiation with the proper three-dimensional (3-D) structure is critical for cells to carry out their cellular functions in tissues. Odontoblasts derived from neural crest cells are elongated and polarized with the cell process, which is decisive for one directional tubular dentin formation. We have observed the formation of dentinal tubule-like structures in the teeth of dogs subjected to electrospun nanofiber in an in vivo direct pulp-capping model [10]. Collectively, it can be speculated that the nanotopography of adhesion substrates induces the 3-D structural differentiation of odontoblasts. Previous studies have suggested that cell adhesion leads to cytoskeletal organization which can be controlled by Rho family GTPases [11]. The Rho family of small GTPases consists of RhoA, Rac1, and Cdc42. Among them, Cdc42 induces the formation of filopodia, while Rac1 and RhoA promote the formation of lamellipodia and actin stress fiber, respectively [[12], [13], [14]]. In this study, we investigated the regeneration of the dentinal tubule structure in a pulp-exposure animal model and the underlying mechanisms on the 3-D polarization of odontoblasts induced by the nanofibrous topography. Methodologically, we implanted electrospun poly(ε-caprolactone) nanofibers on the mechanically exposed pulp of the 1st molar of rats. Additionally, we cultured pre-odontoblast MDPC-23 cells on electrospun polystyrene nanofibers (FIBER) and undertook z-axis imaging to examine whether the cells grown on the FIBER could be induced to having the 3-D polarization of odontoblasts for ortho-dentin regeneration by the nanotopographic substrates. Furthermore, we investigated the underlying molecular events by which the nanofibrous topography-induced Wnt5a regulates the rearrangement of the actin-cytoskeleton and cell shapes.
A polystyrene nano-fibrous matrix was fabricated by electrospinning. Briefly, the polystyrene beads (Sigma, St. Louis, MO) were dissolved in dimethylformamide (DMF) (Sigma), and a 12% (w/v) polystyrene solution added to a 10-mm syringe with a 30G stainless syringe needle. The electric potential and the collector distance were optimized to 30 kV and 20 cm, respectively. The polystyrene nanofiber was collected on a metallic rotating drum. After the electrospinning, the DMF was spread on the dish, and a polystyrene nanofiber sheet was placed over it. The FIBER fixed onto the dish was dried in a vacuum chamber at room temperature. Subsequently, the FIBER on the dish was washed with ethanol (70%) and then dried on a clean bench with UV light overnight at room temperature.
For in vivo studies, the mandibular first molars of 6 rats (8-week-old) were operated on. Control and experimental groups were randomly allocated to each animal to avoid possible biases resulting from individual characteristics and pulpal conditions could be eliminated. Investigators performing the animal operation and sampling were not blinded while the investigators that accessed the histological analysis were blinded. All results using animals followed the protocols approved by the Institutional Animal Care and Use Committee of Seoul National University (SNU-200512-3-5). Tooth preparation was performed as previously described [15]. Briefly, after disinfecting the mandibular first molars with 0.5% chlorhexidine, pinpoint pulp exposure was made on the occlusal surface of the mandibular first molars using a round bur (φ = 0.6 mm). After sufficient irrigation, the exposed sites were dried with a cotton pellet. The cavities were either untreated or treated with poly(ε-caprolactone) nanofibers. After treatment, the cavities were then filled with a calcium silicate cement (ProRoot) and a glass ionomer cement (Fuji II LC; GC America Inc.). The Unsealed group was unable to maintain FIBER during the recovery period and oral microbiological contamination altered the results. Despite the sealing material's ability to form restorative dentin, FIBER alone cannot suggest results. The samples were cut into blocks after 4 weeks following surgery, fixed in 4% paraformaldehyde, and kept for one day at 4 °C. Decalcification was performed in 10% EDTA (pH 6), and then the samples were embedded in paraffin. Serial sections were used for H&E stained (5-μm-thick).
The MDPC-23 mice odontoblast-like cell line was used in this study [16]. The cells were maintained in Dulbecco's modified Eagle's medium (DMEM) (GibcoBRL, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) (GibcoBRL, Carlsbad, CA). For each samples, the cells were seeded on non-patterned polystyrene tissue culture plates (TCP) (BD Falcon, Franklin Lakes, NJ) as a control or on the FIBER seeded of 5 × 105 cells for each 60-mm dish. After 12-h form cell seeding, the cells were cultured in a differentiation condition (added with 10 mM β-glycerophosphate and 50 μg/ml ascorbic acid (Sigma) in growth medium).
The polystyrene nanofibers were washed with PBS and fixed using 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4). After freeze-drying, each specimen was dehydrated using serial dipping in increasing concentrations of ethanol, after which each underwent critical point drying. The samples were sputter-coated with a gold-palladium mixture and observed using scanning electron microscopy (SEM) (FE-SEM Hitachi S-4700, Japan) at 12 kV.
The calcium deposition outcomes of the dental pulp cells on different substrates were analyzed by Alizarin Red S staining. MDPC-23 cells were seeded on six-well cell culture plates without and with the attached FIBER. The next day, the media were changed, and differentiation was allowed to proceed. After seven days of differentiation, the samples were washed three times with cold PBS, fixed with chilled ethanol (70%) for 1 h, washed with deionized water, and stained with 40 mM Alizarin red S (pH 4.2) (Sigma) for 30 min at room temperature. After staining, samples were rinsed with deionized water. Images were captured of the stained samples showing the deposition of calcium.
To determine the morphological alterations, the cells were seeded on the FIBER or TCP at a density of 1 × 104 cells per cm2 in 24-well cell culture plates. After incubated the cells 12-h, cells were washed and cultured in differentiation media for 10 days. The cells were washed with PBS and then treated with a freshly prepared working solution of the CellMask™ (Molecular Probes, Eugene, OR) in a warm serum-free media from a 1000X concentrated CellMask™ stain solution for 60 min in an incubator at 37 °C. The cells were washed with PBS and fixed with 4% formaldehyde for 10-min at room temperature, after which they were washed again three times with 1X PBS. The cells were treated with Alexa Fluor™ 488 Phalloidin (Molecular Probes, Eugene, OR) at a dilution of 1:100 for 30 min at room temperature in the dark. After washing the samples with PBS, the samples were mounted using a mounting medium with 4′,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA) to identify the nuclei. Confocal laser scanning microscopy images were taken to determine the cell morphology (Zeiss, LSM 700). The images were taken from the bottom to the top of the cells at 3 μm intervals. The experiments were independently repeated three times at least.
For the cell height and nuclei polarization measurements, the cells were cultured on TCP and FIBER for four days. To determine the cell height, the bottom of the cells was taken at a zero point and adjusted with confocal laser scanning microscopy at a 0.16 μm interval of each stacked section. The number of stacked intervals from bottom to top for each cell was measured. At the same time, the position of the nucleus was determined at the upper part of the cell based on the height of each respective cell. For quantitative measurements, 100 cells from each group were used for an analysis in each individual experiment, and each experiment was performed at least three times.
The cells were cultured on TCP and FIBER in the differentiation medium. Total RNAs were extracted using RNA isoplus reagents (Takara, Kyoto, Japan), and the cDNA was synthesized using a PrimeScripts RT reagent kit (Takara, Kyoto, Japan) according to the manufacturer's instructions. RT-qPCR was performed on a real-time PCR system using SYBR® Premix Ex TaqTM (Takara, Kyoto, Japan) according to the protocol described in the kit. The relative expression of each target gene transcript was normalized using the level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Each experiment was performed for three biological replicates. The primers are listed in Supplementary Fig. S11.
The cell lysates were analyzed by western blot to detect the specific proteins. The cell lysates were lysed using a lysis buffer consist of 10 mM Tris-Cl (pH 7.5), 150 mM NaCl, 1 mM EDTA (pH 8.0), 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 50 mM NaF, 0.2 mM Na3VO4, a phosphatase inhibitor, and a protease inhibitor cocktail tablet (Roche, Basel, Switzerland). The running gels were used for 10–12% SDS–PAGE and then transferred onto a polyvinylidene difluoride membrane. The membrane blocking performed by 5% skim milk and incubated with each antibody. Immunoreactivity was developed by an enhanced chemiluminescence (ECL) reagent. For the control, anti-β-actin mIgG antibody was used.
ON-TARGET plus SMART pool small interfering RNAs (siRNA) were used to knock down the expression of Wnt5a, Ror2, RhoA, or Cdc42. A scrambled siRNA was used as a negative control. MDPC23 cells were seeded for 5 × 105 cells in 60-mm culture dish with or without FIBER and transfected using transfection reagent with each specific siRNA in accordance with the manufacturer's guidance (Dharmacon, Lafayette, CO). The cells were harvested after 36 h after the transfection.
The activity of small GTPases was determined using a pull-down assay kit following the protocol described by the manufacturer (Cytoskeleton, Denver, CO). Briefly, the cells were cultured on TCP and FIBER to induce cell differentiation. On the fourth day of differentiation, the cells were harvested and lysed using a protease inhibitor in cell lysis buffer. The lysates were centrifuged for 1 min at 14,000 rpm at 4 °C. Protein quantification measurements were taken using a Pierce BCA assay kit (Thermo Scientific, Rockford, IL), and for each sample, 20 μg of protein lysate were used for the total specific small GTPase protein determination. Equal amounts of protein lysate from the remaining supernatants were incubated with Rhotekin-RBD beads for the RhoA activation assay, and equal amounts of protein lysate were mixed with PAK-PBD beads for the Cdc42 activation assay. The samples were incubated on a rotator for 1 h at 4 °C and centrifuged for 1 min at 3000 rpm, after which the supernatant was removed. After washing the beads once with 300 μl each with a wash buffer, the samples were centrifuged at 4000 rpm at 4 °C for 3 min, and the supernatant was removed, and 30 μl of 2x sample buffer were added to each sample. We then thoroughly resuspended the beads. The samples were boiled for 2 min and assessed by western blot analysis.
All data are statically analyzed as the average and standard deviation (SD). The statistical methods using ANOVA or Student's t-test for described significancy. Differences between groups were considered significant if the p value was less than 0.05.
An electrospun nanofibrous scaffold with a dimeter in the homogeneous range of 200–300 nm was fabricated from a poly(ε-caprolactone) solution (Fig. 1A and B) [4,6]. To examine the effects of the nanofibrous topography on the regeneration of tubular dentin in vivo, pulp exposure was mechanically generated in rat molars which were divided randomly into two groups. After four weeks, amorphous calcified tissue formation was observed in the pulp cavity of the conventionally treated control group. In contrast, newly formed dentin retaining the physiologic tubule structure was found beneath the defect in the FIBER-implanted group (Fig. 1C). As shown in Fig. 1D, immunohistochemistry showed that a strong expression of dentin sialoprotein (DSP), which is a molecular marker of odontoblast differentiation, was observed along with the dentinal tubules in the FIBER group. Notably, the DSP-expressing cell bodies were present under the newly formed dentin, and the cytoplasm extended to the dentinal tubules in the FIBER group, while DSP-expressing odontoblasts were mainly found below or between the amorphous calcium structures in the control group. The length from the pulp to the tip of newly formed tubule was significantly increased in the histological images of the FIBER group (Fig. 1E). Given that elongation and 3-D polarization proceed along with odontoblast differentiation and the asymmetric disposition of organelles and cytoskeletal arrangements are essential for normal dentin formation [17], this figure indicated that the nanofibrous scaffold supported the 3-D structural differentiation of odontoblasts as well as their marker gene expression.
The dental tubule structure of newly formed dentin in the in vivo pulp exposure model suggests that the fibrous topography induces the apical growth of odontoblasts in the FIBER group. It was investigated whether the intracellular organization is affected by the nanotopographic surface to which the cells adhere. Cell polarization and cellular process formation are essential changes during odontoblast differentiation [18,19]. We observed that the MDPC23 pre-odontoblasts grown on the electrospun FIBER only showed significant increases in polarity and branching (Supplementary Fig. S1), but TCP did not. To confirm whether this phenotype was linked to dentin matrix formation, MDPC23 cells cultured on FIBER exhibited intense alizarin-red staining, indicating that FIBER promoted odontoblast differentiation, consistent with our previous report (Fig. 2A) [18,19]. In the TCP group, actin fibers were arranged in a flat configuration throughout the cytosol and formed stress fibers, while condensed and slender antenna-like actin structures were observed in the FIBER group (Fig. 2B). Physiological odontoblasts show an increase in the cell height and asymmetrical displacement of the nucleus when the cytoplasm is apico-basal polarized [20,21]. As shown in Fig. 2C, we observed that the cell height increased significantly (1.8-fold at day 7) when cultured on FIBER and increased further with the passage of the incubation period. The number of cells that exhibited nuclei at the upper part of the cells was also increased significantly in the cells on FIBER compared to those on the control TCP (Fig. 2D). In the z-axis images from confocal laser microscopy, it was also observed that the cells on the FIBER group appeared cylindrical and had an asymmetric nuclear upper location compared to those on TCP. The height of MDPC23 cells grown on FIBER was significantly increased on the z-axis. In the early stage of differentiation (day 4), the z-axis growth increased by more than 1.6 times compared to TCP (about 15 and 24 μm for TCP and FIBER, respectively), and at the advanced stage of differentiation (day 10), it was increased by more than 1.8 times (about 15 and 27 μm for TCP and FIBER, respectively) (Fig. 2E and F). Filopodia leads to the adoption of a polarized morphology of the cells [22]. Odontoblasts are connected to other cells via junctional complexes, and increased expressions of zona occludens-1 (ZO-1) and claudin-1 have been observed in mature human odontoblasts [23]. Consistent with the cell elongation, the expression levels of ZO-1 and claudin-1 were significantly increased in the cells on FIBER compared to those on TCP (Supplementary Fig. S2). These results indicate that the nanofibrous topography-induced odontoblast differentiation is accompanied by apico-basal polarized, 3-D structural changes.
It has been reported that wnt5a can regulates the growth and patterning of teeth during animal development [24]. However, it has not been specifically studied how Wnt5a works for tooth development. In this study, pre-odontoblasts grown on FIBER increased Wnt5a expression, as confirmed by RT-qPCR and western blot analyses after four days of odontoblastic differentiation (Fig. 3A and B). To address the effect of Wnt5a on 3-D apico-basal polarization, the cells were treated with recombinant Wnt5a (rWNT5a) protein or knockdown (KD) siRNA. The cell height on FIBER was significantly increased compared to that on TCP and further increased by the rWNT5a treatment. Consistently, the KD of the Wnt5a expression through siRNA (si-Wnt5a) successfully reduced the Wnt5a expression and induced a significant decrease in the cell height in the FIBER group (Fig. 3C and Supplementary Figs. S3, S4, and S5). Nuclei polarization to the upper part of the cells was increased in cells on FIBER, and this polarization was further increased by the rWNT5a treatment and decreased by KD against Wnt5a (Fig. 3D). When the TCP group was treated with rWNT5a, actin stress fibers became stronger, and the cell height did not change notably (Fig. 3E). When the FIBER group was treated with rWNT5a, the formation of filopodia was significantly increased. Additionally, repression of Wnt5a reduced the cell height along with the dissociation of actin fibers (Fig. 3F). These results suggest that Wnt5a would be closely related to the induction of the 3-D structural changes and actin reorganization. Molecular markers of planar cell polarization such as Vangl1, Scrb1, Prickle1 and Ptk7 [[25], [26], [27]] were increased by the rWnt5a treatment, and the nanotopographic structure exacerbated these changes (Supplementary Fig. S10). When the TCP group was treated with rWNT5a, the expression of ZO-1 was mostly located in the perinuclear area. Meanwhile, the ZO-1 in the cells on FIBER further spread toward cell membranes with rWNT5a treatment, evidence by cytoplasmic colocalization of actin filament and ZO-1(Fig. 3G). The expression levels of ZO-1 and Claudin-1 was increased further with the rWNT5a treatment in cells on FIBER compared to those in TCP and significantly decreased with the KD of Wnt5a (Fig. 3H).
Ror2, an orphan tyrosine kinase, is a transmembrane receptor known to mediate Wnt5a-initiated cell migration and filopodia formation [28,29]. To verify whether Ror2 is involved in Wnt5a-induced polarized alterations and actin cytoskeleton reorganization, we knocked down Ror2 gene expression using the siRNA of Ror2 (si-Ror2). The cell height was decreased in the Ror2 KD group compared to the si-control treated cells on FIBER. Moreover, the additional rWNT5a treatment could not rescue the Ror2 knockdown effect on the cell height. Actin stress fiber formation in cells on TCP was absent in si-Ror2 without or with the rWNT5a treatment (Supplementary Fig. S6). It was confirmed a Ror2 association with Wnt5a-induced actin-based filopodia formation in cells on FIBER and demonstrated that filopodia protrusions were completely absent during the Ror2 knockdown in samples both without and with the rWNT5a treatment (Fig. 4A). A quantitative analysis of the cell height and nuclei polarization outcomes also showed the consistency with filopodia formation (Fig. 4B and C). As shown in Fig. 4D and E, the knockdown of Ror2 by siRNA abrogated the expression levels of the tight junction molecules of ZO-1 and claudin-1. Moreover, cells exhibited similar effects on ZO-1 and claudin-1 mRNA expressions after the knockdown of Ror2 even with rWnt5a treatment. It has been shown that irrespective of Wnt5a stimulation, the overexpression of Ror2 can induce filopodia formation by actin reorganization and that the knockdown of Ror2 disrupts the formation of filopodia in HEK293T cells [30]. To knock down the Ror2 gene, we used siRNA against Ror2 and determined the knockdown effect by western blot analysis (Supplementary Fig. S4). These results suggest that Ror2-mediated Wnt5a is involved in 3-D polarized changes, which was confirmed by the inhibition of a functioning noncanonical Wnt signaling pathway.
Rho family small GTPases have been revealed as essential regulators of the actin cytoskeleton system and can be activated by Wnt5a/Ror2 [31]. It has been documented that the Rho family of small GTPases, such as RhoA, can regulate the cell shape through cytoskeleton reorganization to form stress fibers [13]. Another of the small GTPases, Cdc42, has important roles in the organization of filopodia structures and in maintaining cell polarization [13,32]. As shown in Fig. 5A, the cell height and nuclei polarization were rather increased in the cells on FIBER after the knockdown of RhoA (si-RhoA), while the cells on TCP did not show any significant alterations in the cellular height or nuclei polarization. We examined the involvement of Cdc42 in the Wnt5a-induced 3-D polarized alterations and actin-based filopodia formation. The cell height was decreased after the knockdown of Cdc42 (si-Cdc42) in the cells on the nanofibrous matrix, and a similar effect was observed with regard to nuclei polarization. Interestingly, the formation of long actin stress fibers was increased in cells on TCP after Cdc42 knockdown and was further increased with the rWNT5a treatment (Supplementary Fig. S8). During the si-RhoA treatment, the cells on FIBER exhibited an increased filopodia formation, and the cells on TCP exhibited a decreased stress fiber formation (Fig. 5B and C and Supplementary Fig. S7). These alterations of the cell height and nuclei polarization during the knockdown against RhoA were further enhanced in cells on FIBER upon treatment with rWNT5a (Fig. 5D and E). The knockdown effect of Cdc42 on the cell height and nuclei polarization in the cells on FIBER could not be rescued by the rWNT5a treatment (Fig. 5F and G). Our results showed that filopodia formation was completely absent in cells on FIBER after the knockdown of Cdc42 and that the treatment of rWNT5a did not rescue the effect of the cdc42 knockdown in cells on FIBER (Fig. 5H and I).
Small GTPases RhoA and Cdc42 may act contrary to each other because the up-regulation of Cdc42 leads to filopodia formation which is accompanied by RhoA down-regulation [33]. To confirm the pivotal role of the nanotopography during the Wnt5a signaling to Cdc42 or RhoA, we conducted pull-down assays. We observed highly active RhoA in the cells on TCP, for which the level was further increased by the rWNT5a treatment (Fig. 6A), while the active Cdc42 was not detected in the TCP group. In contrast, in the FIBER group, the active Cdc42 was high and further increased by the rWNT5a treatment, while active RhoA was not observed in the FIBER group (Fig. 6B). Additionally, Cdc42 and RhoA were antagonistically activated by a reciprocal reduction mechanism. In the FIBER group, knockdown against RhoA induced an increase in Cdc42 activation. On the other hand, knockdown against Cdc42 increased the amount of activated RhoA in both groups (Fig. 6C and D). The activation of Cdc42 induced the activation of c-Jun N-terminal kinases (JNK), including the MAPK pathway [34]. Interestingly, in the polarized MDPC cells on FIBER, the active Cdc42 increased the phosphorylation of JNK and translocation to nucleus. In addition, the WNT5a treatment further enhanced the activation of JNK signaling (Fig. 6E, Supplementary Fig. S9). Moreover, the expression of planar cell polarity molecules was significantly increased in the FIBER group (Supplementary Fig. S10). Taken together, it was demonstrated that the FIBER-induced Wnt5a regulated the activation of Cdc42 and that this activation was further increased by the rWNT5a treatment. The cells on TCP showed RhoA activation. These results indicate that the nanofibrous topography was closely related to distinctive activation of Cdc42, which prepared cells for the altered responsiveness to Wnt5a and led to 3-D polarization with filopodia, suggesting that the Wnt5a-Ror2-Cdc42 signaling pathway induces the 3-D structural differentiation of odontoblasts (Fig. 6F).
Cells derived from common MSCs and differentiated into distinct lineages exhibit various specific morphologies. Adipocyte cells are fat-laden and have a round shape [35], and osteoblast cells range from cuboidal to elongated depending on the matrix deposition activity [36]. These specific morphologies of cells appear to be optimized for their functions in tissue. Odontoblasts exhibit a polarized and tall columnar shape [37]. With regard to dentin formation, cell polarization and the columnar shape facilitate one directional matrix deposition by odontoblasts. The cell morphologies are believed to arise from alterations in cytoskeletal proteins re-organization [38]. Our study investigated the effects of a nanotopography using a nanofibrous matrix on the morphological changes and on actin reorganization during odontoblast differentiation via Wnt5a and the activated Cdc42. Several signaling pathways and different transcription factors regulate the differentiation of odontoblasts leading to tooth formation during the development process. Among them, bone morphogenetic proteins (BMP), Wnts, Hedgehog family proteins (Hh), and fibroblast growth factors (FGF) have vital roles in tooth development. Alterations of these pathways during odontogenesis can disrupted tooth development. The evidence from animal models indicates that the Wnt signaling pathway has an important role in tooth morphogenesis and that continuous tooth generation is induced by activated Wnt signaling [18,39]. Wnt5a is the most widely studied member of the Wnt family proteins and is critical during the developmental process of various organs. The Wnt5a knock-out mice exhibit small and abnormal patterned teeth with delayed odontoblast differentiation [18]. Wnt5a has gained importance with regard to its role in odontoblast differentiation in mice and humans and is involved in various cellular functions through the regulation of multiple signaling pathways [18,40,41]. It may be possible that cells that exhibit actin-based filopodia protrusions can carry Wnt proteins to the neighboring cells to induce odontoblast differentiation [42]. Moreover, small GTPases are therefore ideally placed to facilitate a significant feature of non-canonical Wnt signaling to induce polarization and actin cytoskeleton reorganization, leading to changes in the cell morphology. An important part of tissue engineering is to create a more favorable ECM microenvironment to guide cell differentiation and tissue regeneration. The topography of synthetic substrates has been shown to guide the differentiation of stem cells [43,44]. The ability to mimic the physical and mechanical properties of the natural ECM is an essential requirement for tissue engineering applications [45,46]. Nanofibrous matrices hold great potential for a mimetic natural ECM, which can modulate cell responses, leading to tissue regeneration [46,47]. Recent progress in nanofabrication techniques are significant and widely used to construct substrates of differing topographies that mimic a fibrillar structure of natural ECM, providing essential support for cellular functions. Using two different materials, in animal experiment were treated with poly(ε-caprolactone) nanofibers, but in in vitro experiment were conducted on polystyrene nanofiber can be suggest the apical growth of odontoblast in vitro and vivo occurred due to the difference in the 3-D topology of nanofibers, not the properties of the materials. We demonstrated that FIBER induced Wnt5a expression followed by odontoblast polarization, which exhibited increased heights and exhibited a columnar appearance with increased actin-based filopodia formations. The junctional ZO-1 and transmembrane claudin-1 expression levels were also increased in cells on FIBER. The actin-based filopodia formation on the nanofibrous matrix was found to be under the control of Cdc42 activation, while stress fiber formation on TCP was connected to the activated RhoA. All these features on FIBER were further augmented by the recombinant Wnt5a protein and abrogated by siRNAs against Wnt5a, Ror2, or Cdc42. Taken together, it was found that the nanofibrous topography-driven altered responsiveness of small GTPase Cdc42 to Wnt5a mediates the three-dimensional polarization of odontoblasts.
Our results demonstrate that nano-topographical cues in the form of a nanofibrous substrate can significantly induce the 3-D structural differentiation of odontoblasts. In a pulp-exposure model, the nanofibrous scaffold supported the 3-D structural differentiation of odontoblasts as well as their marker gene expression. It was confirmed that MDPC-23 pre-odontoblast cells grown on FIBER had a higher cell height, a more asymmetrically located nucleus and cell processes than those on TCP; hence, FIBER provided a more favorable guide for dentin regeneration. Mechanistically, the nanofibrous topography induced the distinctive activation of Cdc42, which prepared cells for the altered responsiveness to Wnt5a and led to a specific differentiated morphology. This study using nanofibrous substrates also implicates that the topography of scaffoldings is closely related to the altered cell responsiveness to their micro-environment and cell adhesion.
Saeed Ur Rahman: Investigation, Methodology, Visualization, Writing – original draft. Woo-Jin Kim: Investigation, Validation, Visualization, Writing – original draft. Shin Hye Chung: Resources. Kyung Mi Woo: Conceptualization, Funding acquisition, Supervision, Writing – review & editing.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. |
|
PMC9647228 | Sarah Ehrenberg,Catherine Walsh Vockley,Paige Heiman,Zineb Ammous,Olivia Wenger,Jerry Vockley,Lina Ghaloul-Gonzalez | Natural history of propionic acidemia in the Amish population | 05-11-2022 | Propionic Acidemia,Amish,Newborn screen,Treatment,Cardiomyopathy,Seizures | Propionic acidemia (PA) in the Amish is caused by a homozygous pathogenic variant (c.1606A>G; p.Asn536Asp) in the PCCB gene. Amish patients can have borderline or normal newborn screening (NBS) results and symptoms can present at any time from early childhood to mid-adulthood. Early diagnosis and initiation of treatment for PA in the non-Amish population improves patient outcomes. Here, we present data from a retrospective chart review of Amish patients diagnosed with PA from three different medical centers in order to document its natural history in the Amish and determine the influence of treatment on outcomes in this population. A total of 38 patients with average current age 19.9 years (range 4y-45y), 57.9% males, were enrolled in the study. Fourteen patients (36.8%) were diagnosed with a positive newborn screening (NBS) while 24 patients (63.2%) had negative or inconclusive NBS or had no record of NBS in their charts. These 24 patients were diagnosed by screening after a family member was diagnosed with PA (14; 58.3%), following a hospitalization for metabolic acidosis (5; 20.8%), hospitalization for seizures (3; 12.5%) or via cord blood (2; 8.3%). The majority of patients were prescribed a protein restricted diet (32; 84.2%), including metabolic formula (29; 76.3%). Most were treated with carnitine (35; 92.1%), biotin (2; 76.3%) and/or Coenzyme Q10 (16; 42.1%). However, treatment adherence varied widely among patients, with 7 (24.1%) of the patients prescribed metabolic formula reportedly nonadherent. Cardiomyopathy was the most prevalent finding (22; 63.2%), followed by developmental delay/intellectual disability (15; 39.5%), long QT (14; 36.8%), seizures (12; 31.6%), failure to thrive (4; 10.5%), and basal ganglia strokes (3; 7.9%). No difference in outcome was obvious for those diagnosed by NBS and treated early with dietary and supplement management, especially for cardiomyopathy. However, this is a limited retrospective observational study. A prospective study with strict documentation of treatment adherence and universal screening for cardiomyopathy and long QT should be conducted to better study the impact of early detection and treatment. Additional treatment options such as liver transplantation and future therapies such as mRNA or gene therapy should be explored in this population. | Natural history of propionic acidemia in the Amish population
Propionic acidemia (PA) in the Amish is caused by a homozygous pathogenic variant (c.1606A>G; p.Asn536Asp) in the PCCB gene. Amish patients can have borderline or normal newborn screening (NBS) results and symptoms can present at any time from early childhood to mid-adulthood. Early diagnosis and initiation of treatment for PA in the non-Amish population improves patient outcomes. Here, we present data from a retrospective chart review of Amish patients diagnosed with PA from three different medical centers in order to document its natural history in the Amish and determine the influence of treatment on outcomes in this population. A total of 38 patients with average current age 19.9 years (range 4y-45y), 57.9% males, were enrolled in the study. Fourteen patients (36.8%) were diagnosed with a positive newborn screening (NBS) while 24 patients (63.2%) had negative or inconclusive NBS or had no record of NBS in their charts. These 24 patients were diagnosed by screening after a family member was diagnosed with PA (14; 58.3%), following a hospitalization for metabolic acidosis (5; 20.8%), hospitalization for seizures (3; 12.5%) or via cord blood (2; 8.3%). The majority of patients were prescribed a protein restricted diet (32; 84.2%), including metabolic formula (29; 76.3%). Most were treated with carnitine (35; 92.1%), biotin (2; 76.3%) and/or Coenzyme Q10 (16; 42.1%). However, treatment adherence varied widely among patients, with 7 (24.1%) of the patients prescribed metabolic formula reportedly nonadherent. Cardiomyopathy was the most prevalent finding (22; 63.2%), followed by developmental delay/intellectual disability (15; 39.5%), long QT (14; 36.8%), seizures (12; 31.6%), failure to thrive (4; 10.5%), and basal ganglia strokes (3; 7.9%). No difference in outcome was obvious for those diagnosed by NBS and treated early with dietary and supplement management, especially for cardiomyopathy. However, this is a limited retrospective observational study. A prospective study with strict documentation of treatment adherence and universal screening for cardiomyopathy and long QT should be conducted to better study the impact of early detection and treatment. Additional treatment options such as liver transplantation and future therapies such as mRNA or gene therapy should be explored in this population.
The distribution of genetic disorders in the Amish differs from the non-Amish population due to a bottleneck effect arising from the migration of a small subset of Anabaptists in Europe to America in the early 18th century due to religious persecution [1]. Propionic acidemia (PA) is caused by pathogenic variants in the PCCA and PCCB genes encoding the alpha and beta subunits of the mitochondrial enzyme propionyl-CoA carboxylase (PCC), leading to absent or reduced enzyme activity [2]. Certain amino acids including isoleucine, valine, threonine, and methionine, in addition to odd chain fatty acids, are important sources of propionyl CoA, which is metabolized by propionyl-CoA carboxylase to D-methylmalonyl-CoA. Defects in PCC lead to toxic metabolite accumulation that results in episodes of intermittent acidosis. In addition, there appears to be a toxic effect on mitochondrial energy metabolism. In the non-Amish population, most patients with PA are compound heterozygotes for variants in PCCB or PCCA [3]. However, Amish patients with PA are homozygous for a missense variant (c.1606A>G; p.Asn536Asp) in the PCCB gene [2,4]. Elevated levels of propionylcarnitine (C3) identified by tandem mass spectrometry as part of newborn screening (NBS) are characteristic of PA. Non-Amish individuals with PA typically present with symptoms in the neonatal period including lethargy, poor feeding, hypotonia, metabolic acidosis and hyperammonemia, often before newborn screening results are available [[5], [6], [7]]. However, Amish patients generally have a less severe, though variable disease, likely because the mutant enzyme has residual activity [2,8]. Newborn screen results usually identify borderline high or normal C3, and patients rarely present with neonatal symptoms. Rather, symptoms can develop at any time from early childhood to mid-adulthood [7]. Clinical findings in PA include cardiomyopathy, cardiac arrhythmias, metabolic decompensation, seizures, basal ganglia strokes, developmental delay and pancreatitis [3,8]. In the Amish population, cardiomyopathy is frequently the presenting symptom. Treatment for PA consists of a modified diet focused on protein restriction, often with supplemental metabolic formula with reduced or free propiogenic amino acids, biotin and carnitine supplementation and occasionally antibiotic treatment to reduce intestinal propiogenic bacteria [2,5,[9], [10], [11]]. Liver transplantation has been shown in many cases to reduce episodes of metabolic decompensation, improve cardiac function in patients with cardiomyopathy, and increase protein tolerance [10,12,13].However, long term comparison of transplanted individuals with non-transplanted patients is lacking in the literature, and some reports have shown that transplanted patients can still develop cardiomyopathy or other complications after liver transplant [10,14].Liver transplantation does not provide complete correction of the metabolic disease and therefore these patients should continue to be managed by metabolic specialist after the transplant. The importance of early diagnosis and treatment of PA for survival has been well documented, but reported long term outcomes have been variable [7,11,15]. Notably, development of cardiomyopathy has been reported in PA patients despite early diagnosis and metabolic management, even with good metabolic control [12,13]. Amish patients with PA in Wisconsin diagnosed and treated early had long-term reduction of neurological sequelae, but no clear change in cardiac complications [7]. The present study is a retrospective examination of medical and laboratory records of Amish patients with PA from Western Pennsylvania, Ohio and Indiana, with a goal of comparing clinical symptoms, progression of disease, and the impact of ongoing management on outcome.
This retrospective study involved the review of charts of Amish patients with PA from Western Pennsylvania, Ohio and Indiana. This study was approved by the University of Pittsburgh Institutional Review Board (protocol #STUDY20050254).
Informed consent to collect documentation of diagnosis, management, and outcomes was obtained from Amish patients with previously diagnosed PA who are followed at three medical centers that are part of the Plain Community Health Consortium (PCHC): UPMC Children's Hospital of Pittsburgh, New Leaf Center, and the Community Health Clinic in western Pennsylvania, Ohio and Indiana, respectively. Parental informed consent was obtained for underaged participants. For patients outside of the UPMC system, charts were shared as either physical copies or electronically on Ecares, a medical record sharing system.
Information abstracted from patient charts included demographic information, initial presentation and age at diagnosis, clinical characteristics, hospitalizations, treatments and outcomes. Laboratory tests captured included NBS results, plasma acylcarnitine profiles, carnitine levels, ammonia, lactate, plasma amino acids, urine organic acid profiles, EKGs, and clinical imaging such as echocardiograms and MRIs of the brain. Treatment parameters documented included intact protein restriction, use of metabolic formula (most with no or limited amounts of isoleucine, valine, threonine, methionine, and odd chain fatty acids), or medication supplementation, and age at treatment initiation. Prespecified diagnoses recognized as sequelae of PA were assessed including cardiomyopathy, arrhythmias, seizures, metabolic strokes and developmental delays/intellectual disabilities.
The goals of the study were to identify common characteristics among Amish patients diagnosed with propionic acidemia and determine the associations of treatment with patient outcome as described in the data collection section.
Data analysis included descriptive statistics to summarize demographic information, positive NBS results, initial presentation, average ages at presentation and diagnosis, average lab results and information on special diets and medications. Percentages of patients with previously specified outcomes were also calculated.
A total of 38 patients were enrolled in the study. The current age of participants at time of data collection was 19.9 years (range 4y-45y), with 57.9% of participants being male.
Of the 38 patients, 22 (57.9%) had a PA NBS reported. Of these, 17 (77.3%) had a presumptive positive result for PA and 5 (22.7%) had negative results. For the 17 patients with presumptive positive NBS, the C3 level on NBS was listed in their chart. The average C3 of these 17 patients was 7.4 ± 1.9 nmol/mL. Normal C3 cutoffs varied depending on the state in which patients received their NBS; for example, the cutoff in Ohio is <5.6 nmol/mL while the cutoff in Wisconsin is <6.92 nmol/mL. One patient had a normal C3/C2 ratio in the setting of a mildly elevated C3 on NBS. A repeat NBS in two patients with slightly elevated C3 was normal. All 3 of these patients were told they likely did not have PA and did not have medical follow up until other family members were diagnosed and they had molecular testing confirmation (Table 1, Fig. 1). Of the 24 patients who were not diagnosed via NBS, 16 participants did not have record of a NBS for PA in their chart, 5 patients had a negative NBS result, and the 3 patients described above were presumed not to have PA despite an initial positive NBS. For these 24 patients, the average age at diagnosis of PA was 5.9 years (3 days to 32 years). The majority (14/24; 58.3%) were diagnosed via testing after a family member was diagnosed. The remaining patients were identified following hospitalization for metabolic acidosis (5/24; 20.8%) or seizures (3/24; 12.5%), or cord blood analysis undertaken because of family history (2/24; 8.3%) (Table 2, Fig. 1). Thirty-three (86.8%) of the 38 participants had record of a molecular study confirming the diagnosis of PA; all were homozygous for the common Amish PCCB (c.1606A>G; p.Asn536Asp) variant (Table 3).
Dietary modifications such as intact protein restriction and intake of metabolic formula were begun at an average age of 4.2 years with implementation ranging from birth to 17 years of age. Thirty-two patients (84.2%) were instructed to follow a protein restricted diet and 29 (76.3%) patients were prescribed metabolic formula. Of patients prescribed metabolic formula, 7 (24.1%) participants were noted to be non-adherent to treatment (Table 4). Most of the metabolic formulas prescribed were the usual commercially available formulas for use in propionic acidemia. Some patients were prescribed other metabolic formulas restricting isoleucine and valine (MSUD formulas such as MSD Jr. and MSD AA formulas) due to availability, then supplemented with oral Leucine so that it was not restricted. The most common medications prescribed for PA were carnitine (92.1%), followed by biotin (76.3%) and coenzyme Q10 (42.1%). Metronidazole was not prescribed for any of the patients at the time of this chart review; however it was subsequently given to one patient due to a glycine level of ∼3 times the normal range and mild cardiomyopathy, despite protein restriction and formula supplementation (Table 5).
Cardiomyopathy was the most prevalent symptom in this cohort (63.2% of patients). Mean age at time of diagnosis of cardiomyopathy was 17.5 years (range 3–38 years), with 37.5% of patients symptomatic at the time of diagnosis. Developmental delay/intellectual disability was the next most common finding, present in 39.5% of patients. Forty-seven percent of these patients had both cognitive and motor delays reported, while 26.7% had only motor and 26.7% had only cognitive delays. History of seizures was present in 31.6% of participants, diagnosed at an average of 1 year of age (range 7 months to 2 years). The most common subtype of seizure was with illness or fever (50.0%), followed by generalized tonic-clonic seizures (41.7%), and focal seizures (8.3%). Long QT was present in 36.8% of patients, diagnosed at an average of 11.3 years (range 8 months to 29 years). Basal ganglia strokes were diagnosed in 7.9% of patients at an average age of 17.2 years (range 7 mo–41 y). Failure to thrive was reported in 4 (10.5%) participants (Table 6). Outcomes for patients were grouped based on types of dietary modifications and medications taken by the patients for the overall cohort as well as subgroups based on the age at diagnosis (Fig. 2, Fig. 3, Fig. 4, Fig. 5). All patients diagnosed by NBS were prescribed a protein restricted diet, and the majority were also prescribed metabolic formula and medications. More patients diagnosed by NBS had no disease sequelae at the time of the chart review than the other two groups; however, patients diagnosed via NBS were on average younger. Patients diagnosed after one year of age had the most variability in treatment and included the lowest number of patients without sequelae. There was no clear pattern in outcome relative to treatment.
Twenty-six patients had an acylcarnitine profile recorded in their chart, all showing an elevated C3 level. Ten of twenty-one patients (47.6%) with a recorded ammonia had mild to moderate elevation (ranging from slightly above normal range to ∼4 times the normal range). Four of sixteen patients (25.5%) with a recorded lactate had an elevated level (ranging from slightly above normal range to ∼3 times the normal range). Thirty-two patients had a carnitine battery in their chart; 7 patients (21.9%) had a low free carnitine and 2 (6.3%) had a low total carnitine. Thirty-two patients had plasma amino acids in their chart, and 27 (84.4%) of these patients had an elevated glycine. Twenty-two patients had urine organic acids reported, with 20 (90.9%) noting a metabolite profile consistent with their PA diagnosis (Table 7).
This study reviewed medical records from Amish patients with a known diagnosis of propionic acidemia in order to characterize the disease course, and assess association of age at diagnosis and treatment implementation with patient outcomes (Table 8). PA is identified by NBS through an elevated C3 level on acylcarnitine profile. In non-Amish patients with neonatal-onset disease, symptoms can arise before NBS results are available [2,16], but Amish individuals typically have milder disease. Thus, the NBS is often normal, or identifies a biochemical abnormality long before symptoms arise [7]. It has been suggested that lowering the C3 cutoff for an abnormal NBS would decrease the number of false negative results [7]. However, this would likely result in an increased number of false-positive results, leading to unnecessary testing, healthcare costs and psychosocial burden for families [17]. Instead, additional targeted screening for individuals of Amish descent has been proposed using molecular testing for the PCCB common Amish variant [7,18]. Over half of the patients in our study had record of a NBS for PA and the majority were positive. For patients without NBS or with a negative NBS, over half were diagnosed with PA after a family member was found to have the condition. A smaller subset of patients had more severe presentations in early childhood such as metabolic decompensation or seizures resulting in hospitalization. This is consistent with other studies showing that PA in the Amish has a variable presentation and is less likely to cause metabolic decompensation in the neonatal period when compared to PA in the non-Amish population [7,19]. Early diagnosis and treatment of inborn errors of metabolism has been shown to have beneficial outcomes for patients [20]. Treatment for PA in the form of an intact protein restricted diet with supplemental metabolic formulas decreases exposure to propiogenic precursors, decreasing episodes of metabolic decompensation and other negative outcomes [2]. In addition, early diagnosis allows care providers, patients, and their families to better prepare for and manage severe metabolic symptoms during periods of otherwise minor illness [21]. Early diagnosis can also facilitate better understanding of the natural history of the disorder, leading to optimization of interventions. Cardiomyopathy has been recognized as a common complication of PA. The age at diagnosis of PA, degree of metabolic control, or degree of residual enzymatic activity do not seem to modify the risk for cardiomyopathy, and cardiomyopathy can occur in patients who have mild to moderate forms of well-controlled PA [12]. Cardiomyopathy can spontaneously resolve or progress to cardiac failure, and lead to sudden death [22]. Cardiac arrhythmias are frequent, including prolonged QTc interval associated with syncope [9,23], and cardiac arrest [24]. The post-partum period also seems to be a high-risk time for women who have PA, with some presenting with cardiomyopathy only after pregnancy. Although there was no difference found in this study for risk for cardiomyopathy relative to age at diagnosis or degree of metabolic control, earlier diagnosis of PA could lead to earlier screening for cardiac dysfunction and subsequent intervention. Per the 2021 management guidelines for PA, EKG and echocardiogram are suggested for monitoring in PA patients. The recommended frequency for echocardiogram is not provided, but suggest for EKGs every 12 months [10]. We suggest for the PA Amish patients a baseline echocardiogram at time of diagnosis and every year thereafter, unless the patient has symptoms of cardiomyopathy. Further follow- up if cardiomyopathy develops will depend on the cardiologist's recommendations. Additionally, adding N-terminal pro–B-type natriuretic peptide (NT-proBNP) to the routine metabolic follow up labs during clinical evaluations could be useful in these patients as it can add value for the prediction of heart failure/ cardiomyopathy [[25], [26], [27]]. Fibroblast growth factor (FGF21) and 2-methylcitrate monitoring also has the potential to predict long term outcomes as seen in previous studies [10]. Liver transplantation has been shown to improve or slow progression of cardiomyopathy secondary to PA, including in individuals with severe heart failure [12,13]. The exact mechanism is not known but could be due to decreased accumulation of toxic metabolites in the heart. In addition, successful liver transplant appears to achieve metabolic stabilization, resulting in fewer hospitalizations, less dietary restriction, and improved linear growth [28,29]. There have, however, been reports of new or recurrent cardiomyopathy following liver transplantation for PA [30,31], though it is unclear if cardiomyopathy is secondary to the presence of a second genetic condition in these individuals or if it is a result of immunosuppressant medications. This is especially important to keep in mind in the Amish since other genetic causes of cardiomyopathy and arrythmia are well described in them [32]. Therefore, gene panel testing for cardiomyopathy/arrythmia in Amish patients diagnosed with PA is highly recommended upon diagnosis with propionic acidemia. In fact, one of the patients in this study had worsening heart failure despite PA management and was being evaluated for liver transplant. He underwent gene panel testing for cardiomyopathy and was found to have a truncating heterozygous pathogenic variant in the TTN gene (c.59693G>A; p.Trp1989*), encoding the sarcomeric protein titin. This changed his management plan with evaluation for heart instead of the liver transplant. Early diagnosis and treatment in patients with neonatal-onset PA decrease risk of metabolic decompensations and basal ganglia strokes. Although Amish individuals with PA are less likely to present with severe metabolic decompensations early in life, patients in this study were diagnosed with episodic metabolic acidosis, seizures, developmental delays/intellectual disabilities, and metabolic strokes. Prediction of which patients will have recurrent metabolic decompensations and associated sequelae is not possible, therefore it is important to diagnose and begin treatment for all patients as early as possible. Without consistent dietary modifications/adherence and adherence to medication regimen, it is unclear if current conservative treatment significantly improves outcomes in this population, especially cardiomyopathy and arrythmias. A prospective research study including strict dietary control, medication monitoring, close monitoring of metabolic labs and predictive biomarkers, follow up echocardiograms and EKGs throughout the course of the disease and structural neurodevelopmental assessment for all patients, in addition to investigating other secondary genetic factors, is required to better understand the natural history of PA disease in patients with the Amish variant. Additional existing therapies such as liver transplantation or new emerging therapies such as mRNA or gene therapy could be considered in this population. Limitations to this study are inherent to the nature of chart reviews. The amount and type of information present in patient charts varied widely depending on how often the patient utilized the healthcare system. Although the majority of patients were prescribed some sort of dietary modification such as protein restriction or metabolic formula, adherence to these recommendations often was not recorded in charts. In addition, adherence to other therapeutic recommendations, especially screening for cardiac complications, was inconsistent, with many apparently receiving no screening. Given that nearly half of the patients with cardiomyopathy were asymptomatic at time of diagnosis, it is possible that consistent monitoring and screening would identify a larger proportion of patients with cardiomyopathy. Lastly, screening Amish couples for the common Amish PA variant is recommended given the possibility of missing PA diagnosis on the NBS in this population.
The authors have no competing interests to declare.
LGG is funded in part by the 10.13039/100000051National Human Genome Research Institute (NHGRI) grant #1K08 HG010490, a component of the National Institutes of Health (NIH). The contents of this manuscript are solely the responsibility of the authors and do not necessarily represent the official view of NHGRI/NIH.
Sarah Ehrenberg: Data curation, Writing - original draft. Catherine Walsh Vockley: Data curation, Project administration, Supervision, Writing - review & editing. Paige Heiman: Data curation. Zineb Ammous: Writing – review & editing. Olivia Wenger: Writing – review & editing. Jerry Vockley: Writing – review & editing. Lina Ghaloul-Gonzalez: Conceptualization, Methodology, Supervision, Writing - review & editing. |
|
PMC9647234 | Tjalf Ziemssen,Marie Groth,Benedict Rauser,Tobias Bopp | Assessing the immune response to SARS-CoV-2 mRNA vaccines in siponimod-treated patients: a nonrandomized controlled clinical trial (AMA-VACC) | 08-11-2022 | COVID-19 vaccination,disease-modifying therapy,neutralizing antibodies,secondary progressive multiple sclerosis,T-cell response | Background: Systematic data are lacking on the immune response toward SARS-CoV-2 mRNA vaccination in SPMS patients on disease-modifying therapies (DMTs). Objective: The AMA-VACC clinical trial was designed to characterize immune responses to SARS-CoV-2 mRNA vaccines in siponimod-treated SPMS patients. Design: AMA-VACC is an ongoing three-cohort, multicenter, open-label, prospective clinical study. Methods: The study included patients at risk for SPMS or patients with SPMS diagnosis. Patients received SARS-CoV-2 mRNA vaccine as part of their clinical routine during ongoing siponimod treatment (cohort 1), during siponimod treatment interruption (cohort 2), or while on dimethyl fumarate, glatiramer acetate, beta-interferons, teriflunomide, or no current therapy (cohort 3). SARS-CoV-2-specific neutralizing antibodies and T-cell responses were measured 1 week and 1 month after the second dose of vaccination. Results: In total, 17 patients, 4 patients, and 20 patients were recruited into cohorts 1, 2, and 3, respectively. The primary endpoint of seroconversion for SARS-CoV-2-neutralizing antibodies at week 1 was reached by 52.9%, 75.0%, and 90.0% of patients in cohorts 1, 2, and 3, respectively. For 64.7% of patients in cohort 1, all patients in cohort 2, and 95% of patients in cohort 3, seroconversion was observed at either week 1 or month 1 or both time points. After 1 week, 71.4% of cohort 1, 75.0% of cohort 2, and 85.0% of cohort 3 were positive for either SARS-CoV-2-neutralizing antibodies or SARS-CoV-2-specific T-cells or both. After 1 month, the rates were 56.3%, 100.0%, and 95.0%, respectively. Conclusion: The study shows that the majority of siponimod patients mount humoral and cellular immune response under continuous siponimod treatment. The data do not sufficiently support interruption of treatment for the purpose of vaccination. Registration: EU Clinical Trials Register: EudraCT 2020-005752-38 (www.clinicaltrialsregister.eu); ClinicalTrials.gov: NCT04792567 (https://clinicaltrials.gov). | Assessing the immune response to SARS-CoV-2 mRNA vaccines in siponimod-treated patients: a nonrandomized controlled clinical trial (AMA-VACC)
Systematic data are lacking on the immune response toward SARS-CoV-2 mRNA vaccination in SPMS patients on disease-modifying therapies (DMTs).
The AMA-VACC clinical trial was designed to characterize immune responses to SARS-CoV-2 mRNA vaccines in siponimod-treated SPMS patients.
AMA-VACC is an ongoing three-cohort, multicenter, open-label, prospective clinical study.
The study included patients at risk for SPMS or patients with SPMS diagnosis. Patients received SARS-CoV-2 mRNA vaccine as part of their clinical routine during ongoing siponimod treatment (cohort 1), during siponimod treatment interruption (cohort 2), or while on dimethyl fumarate, glatiramer acetate, beta-interferons, teriflunomide, or no current therapy (cohort 3). SARS-CoV-2-specific neutralizing antibodies and T-cell responses were measured 1 week and 1 month after the second dose of vaccination.
In total, 17 patients, 4 patients, and 20 patients were recruited into cohorts 1, 2, and 3, respectively. The primary endpoint of seroconversion for SARS-CoV-2-neutralizing antibodies at week 1 was reached by 52.9%, 75.0%, and 90.0% of patients in cohorts 1, 2, and 3, respectively. For 64.7% of patients in cohort 1, all patients in cohort 2, and 95% of patients in cohort 3, seroconversion was observed at either week 1 or month 1 or both time points. After 1 week, 71.4% of cohort 1, 75.0% of cohort 2, and 85.0% of cohort 3 were positive for either SARS-CoV-2-neutralizing antibodies or SARS-CoV-2-specific T-cells or both. After 1 month, the rates were 56.3%, 100.0%, and 95.0%, respectively.
The study shows that the majority of siponimod patients mount humoral and cellular immune response under continuous siponimod treatment. The data do not sufficiently support interruption of treatment for the purpose of vaccination.
EU Clinical Trials Register: EudraCT 2020-005752-38 (www.clinicaltrialsregister.eu); ClinicalTrials.gov: NCT04792567 (https://clinicaltrials.gov).
Limited data are available on the efficacy of SARS-CoV-2 mRNA vaccines in patients with secondary progressive multiple sclerosis (SPMS) receiving disease-modifying therapies (DMTs). Some evidence suggests an impaired antibody response in DMT-treated patients. The response to SARS-CoV-2 infection is characterized by the development of IgG, IgA, and IgM antibodies, including neutralizing antibodies (NAbs) able to block viral entry. In addition, T-cells play a central role. Major histocompatibility complex (MHC) I class presentation of intracellular viral peptides activates CD8+ cytotoxic T-cells. Upon destruction of the host cells, viral peptides are released and presented by MHC class II antigen-presenting cells. This triggers CD4+ T-cell expansion and differentiation to T-helper 1 (Th1) and T-helper 2 (Th2) cells. CD4 + Th2-cells then stimulate B-cells to produce virus-specific antibodies, while CD4+ Th1-cells activate T-cell-mediated antiviral response. SARS-CoV-2 mRNA vaccines mimic the natural antiviral reaction, including initial cellular mRNA uptake and subsequent production, presentation, and release of the viral spike protein. The mRNA vaccine by Moderna (mRNA-1273) and the BioNTech/Pfizer mRNA vaccine (BNT162b2) both encode the full-length spike protein that has been stabilized in its prefusion conformation (S-2P). BNT162b2 and mRNA-1273 have both demonstrated efficacy in preventing COVID-19 in persons not receiving immunosuppressive or immunomodulating medications. Both have been shown to induce anti-spike protein antibodies and NAbs in all patients after the second dose of vaccination as well as Th1-based CD4+ and CD8+ T-cell reactivity against the spike protein. The initial response to mRNA vaccines therefore involves CD8+ cytotoxic T-cells, while Th1- and Th2-cells mediate the secondary response including antibody production. It is therefore essential to investigate both humoral and cellular immune reactivity toward mRNA vaccines in patients receiving DMTs like siponimod, which inhibits S1P1 and S1P5 receptors on lymphocytes inducing their retention in lymph nodes. The summary of product characteristics (SmPC) for siponimod suggests considering temporary treatment interruption for vaccination. Protein-based influenza vaccines have proven adequate reactivity in siponimod patients. However, it remains unclear whether siponimod impacts the immune response to the novel mRNA-based vaccines and whether treatment interruption is necessary in this particularly vulnerable patient population. This study was designed to characterize the cellular and humoral immune response to SARS-CoV-2 mRNA vaccines in siponimod-treated SPMS patients with and without treatment interruption, and in a control group receiving dimethyl fumarate, glatiramer acetate, beta-interferons, teriflunomide, or no DMT. With these data, we aim to offer a guidance to treating physicians and patients for the coordination of multiple sclerosis (MS) therapy and vaccination.
[An Open-label Multicenter Study to Assess Response to SARS-CoV-2 modRNA VACCines in Participants With Secondary Progressive Multiple Sclerosis Treated With Mayzent (Siponimod)] (AMA-VACC) is an ongoing three-cohort, multicenter, open-label, prospective clinical study (EudraCT 2020-005752-38; NCT04792567) over 6 months in Germany. The study population consists of patients with SPMS diagnosis and patients with relapsing-remitting multiple sclerosis (RRMS) at risk of developing SPMS. No criteria for RRMS at risk of SPMS transition were applied. Selection of patients of this category was at the discretion of the physician. Patients had to be eligible and planning to receive an SARS-CoV-2 mRNA vaccine as part of clinical routine. Patients with acute [assessed by polymerase chain reaction (PCR)] or previous SARS-CoV-2 infection (assessed by IgA levels ⩾0.8 index and IgG levels ⩾50 AU/ml) at screening were excluded. SARS-CoV-2 vaccinations were administered according to the German vaccination guidance at dedicated sites outside this study. Booster vaccinations are allowed as part of the clinical routine at the physician’s discretion. Patients are treated with siponimod, glatiramer acetate, dimethyl fumarate, interferons, or teriflunomide or currently receive no DMT as part of their clinical routine. The first cohort consists of participants vaccinated during ongoing siponimod treatment. The second cohort includes participants interrupting their siponimod therapy for vaccination. Physician and patient decided upon the preferred option, which are both according to the siponimod SmPC. The third cohort (control group) receives vaccination while on dimethyl fumarate, glatiramer acetate, beta-interferons, teriflunomide, or no DMT.
The trial is conducted in accordance with the International Conference on Harmonisation guidelines for Good Clinical Practice and the principles of the Declaration of Helsinki. The protocol was approved by the ethics committee ‘Technische Universität Dresden’ (AMG ff-EK-34012021). All patients or their legal representatives provided written informed consent before commencing trial-related procedures.
The primary endpoint is the proportion of participants achieving seroconversion as defined by detection of SARS-CoV-2 NAbs 1 week after the second dose of vaccine. Secondary endpoints are SARS-CoV-2 serum NAb levels and SARS-CoV-2-specific T-cell reactivity. NAbs are analyzed utilizing the cPassTMSARS-CoV-2 Neutralization Antibody Detection Kit from GenScriptUSA Inc (L00847) and the assay specific cut-off was used for interpretation. SARS-CoV-2-specific T-cell reactivity is assessed ex vivo by enzyme-linked immunosorbent spot (ELISpot) assays measuring the release of interleukin-2 (IL-2) or interferon-gamma (IFN-γ) from isolated peripheral blood mononuclear cells (PBMCs, 2 × 105) upon antigen stimulation with SARS-CoV-2 peptide mix. The CoV-iSpot Interferon-γ plus Interleukin-2 (ELSP 7010 strip format) from GenID is used. Cross-reactivity with other coronaviruses is assessed using the same ELISpot assay after PBMC stimulation with a homologous coronavirus family peptide mix (pan-corona peptide mix). Assessments are performed at 1 week, 1 month, and 6 months after the second dose of the vaccination cycle. If booster vaccinations are performed, an additional study visit is planned 1 month after this booster vaccination. A final follow-up call to assess the occurrence of COVID-19 infections is intended 12 months after the second dose of the vaccination cycle.
The results of a pre-planned interim analysis, scheduled after all participants have completed the study visit 1 week after the second vaccination, are presented (data cut-off: 31 January 2022). This analysis constitutes the primary analysis of the study. The primary endpoint results, month 1 seroconversion data and secondary data on T-cell reactivity at week 1 and month 1 are presented along with safety data. Month 6 data and antibody levels will be presented separately with the final study results. No formal statistical testing was applied. A sample size of 20 participants per arm was selected based on the need for early availability of results and the feasibility to recruit sufficient participants. The number and the proportion of participants per group achieving seroconversion were analyzed descriptively and presented as frequencies and percentages with a 95% confidence interval (exact Clopper–Pearson). All secondary endpoints were analyzed descriptively and presented as frequencies and percentages, mean and standard deviation (SD), or median and range. Statistical analyses were performed with SAS version 9.2.
In total, 41 MS patients were enrolled at 10 sites in Germany from 19 April 2021 to 4 August 2021, no screening failures were reported. Of these, 17 patients, 4 patients, and 20 patients were recruited into cohorts 1, 2, and 3, respectively. All patients were tested negative for previous or acute SARS-CoV-2 infection by IgA/IgG assessment and PCR. All patients (100%) had completed week 1 visit, and 40 patients (97.6%) had completed month 1 visit at the time of the interim analysis. For one patient, the assessment at month 1 was not performed. Patient characteristics per cohort for this first interim analysis are shown in Table 1. Briefly, median age of participants was 51–56 years and MS history was 9–18 years, with higher age and longer disease history in the siponimod cohorts (cohorts 1 and 2) (Table 1). The mean (±SD) duration of interruption of siponimod for the purpose of vaccination in cohort 2 (n = 4) was 76.7 ± 15.0 days. Siponimod treatment was stopped at a mean 15.3 ± 9.1 days before first vaccination until a mean 29.7 ± 2.9 days after second vaccination, reflecting the suggested treatment break according to the SmPC. The primary endpoint of seroconversion for SARS-CoV-2 NAbs at week 1 was reached by 9 of 17 patients (52.9%) in cohort 1 with continuous siponimod treatment, by 3 of 4 patients (75.0%) in cohort 2 with interrupted siponimod treatment, and by 18 of 20 patients (90.0%) in cohort 3, that is the control group. Seroconversion at month 1 was observed in 9 of 16 patients (56.3%) in cohort 1, in 4 of 4 patients (100.0%) in cohort 2, and in 19 of 20 patients (95.0%) in cohort 3 (Figure 1(a)). In 11 of 17 patients (64.7%) in cohort 1, in 4 of 4 patients (100%) in cohort 2, and in 19 of 20 patients (95.0%) in cohort 3, seroconversion had been observed at either week 1 or month 1 or both time points after full vaccination. One week after vaccination, 7 of 14 patients (50.0%) continuously treated with siponimod, 3 of 4 patients (75.0%) interrupting their siponimod treatment, and 12 of 20 patients (60.0%) in the control group mounted an SARS-CoV-2-specific T-cell response. One month after second vaccination, T-cell reactivity was observed in 0 of 16 patients (0.0%), in 1 of 4 patients (25.0%), and in 14 of 20 patients (70.0%) of cohorts 1, 2, and 3, respectively (Figure 1(b)). Of note, T-cell response could not be assessed in three patients with continued siponimod treatment at the 1-week visit and one patient with continued siponimod treatment at the 1-month visit because of insufficient cell counts after PBMC isolation. Due to its mode of action, siponimod treatment reduces the proportion of CD3+ T-lymphocytes in the blood, which was also observed in this study (Table 2). This resulted in a reduced proportion of CD3+ cells in isolated PBMCs and thus a lower absolute number of T-cells among the 2 × 105 PBMCs plated in ELISpot assays in the continuously treated siponimod cohort. Analysis of combined immune response (development of SARS-CoV-2-specific NAbs or T-cell reactivity or both) showed that 1 week after the second dose of vaccine, 10 of 14 patients in cohort 1 (71.4%), 3 of 4 patients in cohort 2 (75.0%), and 17 of 20 patients in cohort 3 (85.0%) were positive for either humoral or cellular response or both. One month after the second dose of vaccine, 9 of 16 patients in cohort 1 (56.3%), 4 of 4 patients in cohort 2 (100.0%), and 19 of 20 patients in cohort 3 (95.0%) were positive for either humoral or cellular response or both (Figure 1(c)); percentages refer to patients with evaluable assessment]. To analyze the level of cross-reactivity with other coronaviruses, an ELISpot assay measuring IFN-γ or IL-2 secretion of PBMCs after stimulation with a pan-coronavirus peptide mix representative for multiple other coronaviruses was performed. None of the patients with T-cell response against SARS-CoV-2 at week 1 or month 1 in the siponimod cohorts (cohorts 1 and 2) had IFN-γ or IL-2 reactivity against other coronaviruses. For cohort 3, the results showed no IL-2-reactivity but were equivocal for IFN-γ in one patient, suggesting possible previous immune response against other coronaviruses. None of the patients negative for SARS-CoV-2-specific T-cell reactivity at week 1 or month 1 had IFN-γ or IL-2 reactivity against other coronaviruses at screening. Until the cut-off date of this interim analysis, two relapses occurred during the study, both more than 5 months after the last vaccination in cohort 1. No relapses were observed in cohorts 2 and 3. Overall, 25 patients (61.0%) reported adverse events (AEs) during the study. Of these, 20 AEs were related to DMTs or SARS-CoV-2 vaccines (Table 3). One patient from cohort 3 reported serious AEs (acute sinusitis and gastroenteritis rotavirus). One patient from cohort 1 discontinued study medication (siponimod) due to AEs until the cut-off date. No COVID-19 infection was reported until the cut-off date. No deaths occurred.
According to the AMA-VACC interim results, about two-third of patients under continuous siponimod treatment, all patients with siponimod treatment interruption, and almost all patients of the control group developed NAbs within 1 month after vaccination. T-cell response was developed by 50–75% of patients in the different cohorts. Taken together, more than 70% of patients with continuous siponimod treatment, 75% with siponimod interruption, and 85% in the control group developed SARS-CoV-2-specific humoral or cellular response or both as soon as 1 week after full vaccination. SARS-CoV-2-reactivity in all cohorts was neither elicited by prior SARS-CoV-2 infection nor impacted by coronavirus cross-reactivity. Adaptive immune response to SARS-CoV-2 has been shown to involve IgG antibodies as well as CD4+ and CD8+ T-cell reactivity. NAbs, a subset of specific antibodies, have been shown to prevent binding of virus particles to the host cells and interrupt viral entry. They are considered a more stringent correlate of protective immunity than total anti-SARS-CoV-2 antibodies. The presence of NAbs in combination with CD4+ and CD8+ T-cell reactivity have been proposed as suitable predictors of a protective immune response. Therefore, in contrast to other studies, the present study analyzes the development of SARS-CoV-2-specific NAbs instead of total anti-SARS-CoV-2 antibodies. In addition, SARS-Cov-2-specific T-cell reactivity is analyzed. The latter is measured by antigen-stimulated release of IFN-γ and IL-2, which are cytokines released by activated CD4+ and CD8+ T-cells. The results of this study provide important information about whether siponimod-treated patients are able to mount potentially protective immunity after vaccination. It has previously been hypothesized that both humoral and cellular immune responses are functional in patients treated with S1PR modulators as the majority of patients recovers unremarkably from COVID-19. According to a case series, 86% of siponimod patients reported asymptomatic, mild, or moderate infection, and the majority of patients completely recovered or were recovering at the time of data collection. As according to the authors of the case series, their data are subject to potential underreporting of less severe cases typical for post-marketing settings, these results should not be generalized, but at least allow to assume sufficient immunocompetence in siponimod-treated patients. The present findings from AMA-VACC further support this hypothesis and highlight that it is important to consider both humoral and cellular immune responses. T-cells are a prerequisite for B-cell activation and antibody development after vaccination with mRNA vaccines. The development of NAbs suggests initial T-cell reactivity together with functional T-cell–B-cell interaction in the majority of patients even if T-cell responses were not detectable in all patients with NAbs. Up to now, evidence regarding immune reactivity toward SARS-CoV-2 vaccines in siponimod-treated patients was mainly limited to humoral response. A study including 13 patients receiving siponimod and 11 healthy controls detected antibodies in 85% and 100% of patients, respectively. Bar-Or et al. and Conte found similar seroconversion rates for siponimod-treated patients (80% and 88%, respectively). Neutralization activity was not assessed. Results seen with other S1PR modulators are quite diverse with seroconversion rates ranging from 3.8% to 85.7%. Due to differences in pharmacokinetics and pharmacodynamics, the results cannot be transferred to siponimod. Antibody titers in patients receiving siponimod (n = 13) were found to be significantly lower than in healthy controls (n = 11). However, it should be noted that no clinical relevance threshold for antibody titers has yet been established. To what extent lower titers in patients with seroconversion under siponimod might impact the efficacy of the vaccination, if at all, cannot be answered so far. The AMA-VACC study is the first to systematically analyze both humoral and cellular immune responses to SARS-CoV-2 vaccines in patients receiving siponimod. In line with previous publications recommending SARS-CoV-2 vaccination for patients currently receiving DMTs, the present results support vaccination of siponimod-treated patients. The SmPC of siponimod recommends temporary discontinuation 1 week prior until 4 weeks after vaccination. While AMA-VACC shows that SARS-CoV-2 vaccination of siponimod-treated patients induced relevant immune reactivity, the results regarding treatment interruption for vaccination need to be interpreted carefully. It has to be pointed out that patients could be included in either cohort 1 or 2 at the investigators’ discretion. Most physicians prioritized ongoing therapy, probably because of the increased risk of disease activity and progression associated with treatment interruption. This contrasts with a higher immune response rate in the cohort with paused siponimod. However, results from this very small-sized cohort (n = 4) are insufficient to support siponimod interruption. The potential marginal benefit does not outweigh the associated risks. Given that an immune response can be achieved under continued siponimod treatment and a third vaccine dose has meanwhile been recommended to increase vaccination efficiency, treatment interruption is even less favorable. Booster vaccinations are allowed in the AMA-VACC study. The results regarding these booster vaccinations are not available yet and will be described together with the results of the final analysis. Concomitant siponimod use during vaccination is supported by a double-blind, placebo-controlled study with siponimod in healthy volunteers. It was observed that concomitant siponimod was associated with no relevant effect on antibody response to pneumococcal polysaccharide vaccine. Furthermore, most of the patients showed seroconversion for a T-cell-dependent influenza vaccine although titers were lower in comparison with placebo. Furthermore, safety results agreed with previous safety data for both, DMTs and SARS-CoV-2, vaccines. This suggests that vaccination during continuous siponimod treatment is safe and induces SARS-CoV-2-specific immune response. Despite these encouraging results, the present study bears some limitations. First, the study included a small sample size only. Confirmation of the results in further studies is thus necessary. Nevertheless, the results of AMA-VACC allow for the assumption that an immune response toward SARS-CoV-2 mRNA vaccines is elicited in siponimod-treated patients. Second, participants in cohorts 1 and 2 are older and have a longer MS history than participants in cohort 3. Based on recently published data, higher age is negatively correlated with SARS-CoV-2 NAb titers after vaccination and could potentially confound this analysis. Although patients in cohorts 1 and 2 can be considered representative for patients currently treated with siponimod in general clinical practice, the need for earlier diagnosis of SPMS might rejuvenate the typical siponimod population in future. It can be hypothesized that immune response rates in earlier diagnosed and thus younger SPMS patients might be even higher. Noticeably, siponimod treatment in cohort 1 reduced the proportion of CD3+ T-lymphocytes in the blood and the absolute number of plated T-cells in ELISpot assays. This possibly minimized the overall chance of IFN-γ or IL-2 release. This technical problem due to the mode of action of siponimod limits the meaningfulness of this assay in siponimod-treated patients and potentially underestimates the T-cell response in this cohort. Regarding cohort 2, it has to be pointed out that the very small size impacts the meaningfulness of the results and they should not be the basis for rushed treatment decisions. In summary, the results of the pre-planned interim analysis of the AMA-VACC study show that the majority of siponimod-treated patients mounts humoral and cellular immune responses under continuous siponimod therapy. The presented interim analysis data are insufficient to support a general recommendation for an interruption of treatment for the purpose of vaccination. Further results from AMA-VACC on the effect of booster vaccinations, the maintenance of the immune response, and the clinical efficacy regarding COVID-19 in siponimod-treated patients will be published together with the final analysis of the study. |
|
PMC9647246 | 36355184 | Chiara Briani,Sergio Ferrari,Matteo Tagliapietra,Livio Trentin,Andrea Visentin | Vasculitic flare in a patient with anti-myelin-associated glycoprotein (MAG) antibody following mRNA-1273 SARS-CoV-2 vaccine | 10-11-2022 | Vasculitic flare in a patient with anti-myelin-associated glycoprotein (MAG) antibody following mRNA-1273 SARS-CoV-2 vaccine
Dear Sirs, Anti-myelin-associated glycoprotein (MAG) antibody is the most common IgM paraproteinemic neuropathy, characterized by sensory symptoms, gait ataxia, and slowly progressive course. Cryoglobulins, both type I (monoclonal IgMs or IgGs, rarely IgAs) and type II (mixed forms) may also be associated with IgM paraprotein, both of undetermined significance (MGUS) or B-cell malignancies, with a predominance of type II cryoglobulins in Waldenström's macroglobulinemia. The association of anti-MAG antibody neuropathy and cryoglobulins has rarely been described [12, 14] The COVID-19 pandemic and also vaccination against SARS-COV-2 have raised concerns for the worsening of both immune-mediated neuropathies and cryoglobulinemia. Consistently recommendations have been developed by appropriate task forces both for inflammatory neuropathies [6] and for cryoglobulinemic vasculitis [10]. Subsequently the short-term safety of the vaccines has been reported both in inflammatory neuropathies and in cryoglobulinemias. Data from a multicenter Italian study observed post-vaccination vasculitis flares in 5.3% of subjects from a cohort of 416 patients with mixed cryoglobulinemic vasculitis [13]. Despite flares were in line with those observed in other autoimmune diseases [16], patients with purpura or neuropathy seemed at greater risk for symptoms' exacerbation. We report on a patient with long-lasting paucisymptomatic anti-MAG antibody neuropathy who developed a cryoglobulinemic flare and severe neuropathy worsening after the first dose of the mRNA coronavirus disease 2019 vaccination. Sural nerve biopsy documented the vasculitic process.
A 87-year-old woman with a 10-year-history of mild sensory demyelinating neuropathy associated with anti-MAG antibody, complained of mild distal paresthesias at feet that did not affect her gait or functionality, INCAT (Inflammatory Neuropathy Cause and Treatment) Disability Score 0. In her past medical history, she was affected by high blood pressure and hear loss. In May 2021, she underwent the first dose of the mRNA-1273 coronavirus disease 2019 vaccination with rapid worsening of symptoms and occurrence of motor involvement (lower limbs) that required hospitalization. Purpura also occurred. Shortly she became unable to walk and needed wheelchair to walk outdoor (INCAT lower limbs 4) and had trouble in doing zips and buttons (INCAT upper limbs 2). Blood test revealed increased IgM levels (2.96 g/L, normal value 0.4–2.38 g/L), two IgM monoclonal gammopathies (total sum 1.37 g/L), cryoglobulins (2%, monoclonal IgM-type) and increased rheumatoid factor (244Ku/L). Antibodies to MAG were positive 51,404 BTU. Complement was consumed (C3 0.94 g/L, normal range 0.9–1.8 g/L; C4 0.06 g/L, normal range 0.09–0.36 g/L). Levels of anti-MAG antibodies were unchanged after vaccination. On the other hand, cryoglobulinemia was absent before vaccination and present at 2% soon after vaccination. Neurophysiology revealed a severe mixed (demyelinating and axonal) polyneuropathy at four limbs, worse at lower limbs. Sural nerve biopsy (6 months after the vaccination) showed prominent focal axonal loss with rare residual myelinated fibers and axonal degeneration, perivascular epineural infiltrates of mononuclear inflammatory cells also the with presence of hemosiderin deposition. The pathological picture was consistent with a microvasculitic process (Fig. 1). Some residual fiber showed demyelination. Immunofluorescence was negative for IgM deposition and electron microscopy of rare residual fibers did not show widening of myelin lamellae. The patient was treated with steroids with improvement of the active vasculitic skin lesions, but despite intensive physical therapy her gait remained unstable, and she needed bilateral support (walker) also to walk at home, and wheelchair in outdoor space. The patient, who loves painting as hobby, complained of disabling tremor, that prevented her from painting and was never present in the previous years. At neurological evaluation 7 months after the flare onset the patient, who was still in low dose oral steroid therapy (prednisone 10 mg/die), was able to walk only a few steps without assistance with an ataxic and bilateral stepping gait. Strength was reduced distally bilaterally, worse on the right side: tibial anterior 3/5 MRC at the right side, 3.5/5 at left side, extensor hallucis longus and extensor digitorum longus 0/5 at right side, 2/5 at left side. Sensory loss and reduced vibration sense were present up to the knees (0/8 allux, 2/8 ankle, 4/8 knee), deep tendon reflexes were absent at lower limbs. Petechial scars were present in the lower limbs. Total INCAT was 5 (upper limbs 2, lower limbs 3). Steroids were discontinued. The patient underwent bone marrow biopsy that revealed a small clone of k-restricted B lymphocytes CD19+ CD5+ CD11c+ and MYD88 L265P mutation was absent (details on the assessment of MYD88 mutation has been previously reported in [4]. A marginal zone non-Hodgkin lymphoma, which is commonly associated with cryoglobulinemia and sometimes with MAG neuropathy [3] was diagnosed. She underwent therapy with 4 weekly rituximab 375 mg/m2 (from March to April 2022), with prompt benefit. At neurological evaluation at the beginning of April, the patient was able to walk without assistance, although cautiously, she was able to move the fingers of the feet, functionality also improved with decreased need of bilateral support at home. Distal strength had also improved (right anterior tibial 4/5 MRC, left anterior tibial 4.5/5 MRC, right extensor hallucis longus and extensor digitorum longus 2/5 MRC). After treatment serum IgM decreased to 1.47 g/L, sum of monoclonal gammopathies to 0.7 g/L, rheumatoid factor to 58 Ku/L, antibodies anti-MAG titer to 21180 BTU and cryoglobulins disappeared. Complement level normalized (C3 0.94 g/L and C4 0.18 g/L). Five months later (September 2022) the patient showed further improvement. Her gait was possible with no support (although the patients uses the walker to walk outdoor for greater safety), strength fully recovered apart from a mild weakness at right tibial anterior (4.5/5 MRC). No sensory loss was present. Vibration was 0/8 at allux, 2/8 ankle, 5/8 knee, 6/8 index bilaterally. Deep tendon reflexes reappeared at knees. Tremor was absent. INCAT of upper limbs was 0, for lower limbs it was 3. The patient was able to swim during the summer vacation and is undergoing active physical therapy.
Relapse of cryoglobulins vasculitis [9, 15] or other autoimmune diseases [7, 11] after SARS-COV-2 vaccines have already been described. Although autoimmune diseases seem more commonly triggered after adenovirus vectored SARS-CoV-2 vaccines [8], also mRNA vaccines, stimulating the immune system, may worsen autoimmune diseases [7, 9, 11, 15]. However mRNA vaccine for SARS-COV-2 is recommended in cryoglobulinemic vasculitis being the benefit/risk in favor of vaccination [10]. Here we report on a patient with long-lasting paucisymptomatic anti-MAG antibody neuropathy who developer a cryoglobulins flare and likely also a worsening of the underlying autoimmune neuropathy with ataxic gait and onset of disabling upper limbs tremor. The patient quickly responded to steroid therapy, with disappearance of the cutaneous manifestations. However, the gait instability (severe ataxic stepping gait) and tremor did not ameliorate after steroids, and were greatly disabling, limiting patient’s daily activities and autonomy. Rituximab, an anti-CD20 chimeric monoclonal antibody, has been shown to improve cryoglobulins vasculitis [5] and almost half of patients with anti-MAG antibody neuropathy [2]. The findings from sural nerve biopsy confirmed the vasculitic process and did not show IgM deposition or widening of myelin lamellae, characteristics of anti-MAG antibody neuropathy, probably due to the low amount of residual fibers in the sural nerve. Therefore, some of the main pathological characteristics of the anti-MAG neuropathy cannot be found in the sural nerve biopsy, but it was possible to detect the presence of demyelination on the rare teased fibers. Finally, in the sural nerve of our patient, the pathological picture of marked axonal loss secondary due to cryoglobulinemic vasculitis overwhelmed the possible pre-existing alterations due to anti-MAG neuropathy [1]. In our patient, rituximab, probably due to its efficacy on both cryoglobulinemia and anti-MAG neuropathy, greatly improved the clinical picture of the patient ameliorating both the sensory abnormalities, that were fully regained at lower limbs, and motor weakness. Tremor disappeared allowing the patient to resume painting. |
||
PMC9647284 | 36347039 | Guanli Yuan,Yinfeng Liu,Zheng Wang,Xiaotong Wang,Zhuoxiao Han,Xixin Yan,Aihong Meng | PM2.5 activated NLRP3 inflammasome and IL-1β release in MH-S cells by facilitating autophagy via activating Wnt5a | 08-11-2022 | PM2.5,pulmonary inflammation,NLRP3 inflammasome,IL-1β release,autophagy,Wnt5a | Particulate matter 2.5 (PM2.5)-induced pulmonary inflammation is an important issue worldwide. NLRP3 inflammasome activation has been found to be involved in pulmonary inflammation development. However, whether PM2.5 induces pulmonary inflammation by activating the NLRP3 inflammasome has not yet been fully elucidated. This study researched whether PM2.5 induces the NLRP3 inflammasomes activation to trigger pulmonary inflammation. Mice and MH-S cells were exposed to PM2.5, BOX5, and Rapamycin. Hematoxylin and eosin staining was performed on the lung tissues of mice. M1 macrophage marker CD80 expression in the lung tissues of mice and LC3B expression in MH-S cells was detected by immunofluorescence. IL-1β level in the lavage fluid and MH-S cells were detected by enzyme-linked immunosorbent assay. Protein expression was detected by Western blot. Autophagy assay in MH-S cells was performed by LC3B-GFP punctae experiment.PM2.5 exposure induced the lung injury of mice and increased NLRP3, P62, Wnt5a, LC3BII/I, and CD80 expression and IL-1β release in the lung tissues. PM2.5 treatment increased NLRP3, pro-caspase-1, cleaved caspase-1, Pro-IL-1β, Pro-IL-18, P62, LC3BII/I, and Wnt5a expression, IL-1β release, and LC3B-GFP punctae in MH-S cells. However, BOX5 treatment counteracted this effect of PM2.5 on lung tissues of mice and MH-S cells. Rapamycin reversed the effect of BOX5 on PM2.5-induced lung tissues of mice and MH-S cells.PM2.5 activated the NLRP3 inflammasome and IL-1β release in MH-S cells by facilitating the autophagy via activating Wnt5a. The findings of this study provided a new clue for the treatment of pulmonary inflammation caused by PM2.5. | PM2.5 activated NLRP3 inflammasome and IL-1β release in MH-S cells by facilitating autophagy via activating Wnt5a
Particulate matter 2.5 (PM2.5)-induced pulmonary inflammation is an important issue worldwide. NLRP3 inflammasome activation has been found to be involved in pulmonary inflammation development. However, whether PM2.5 induces pulmonary inflammation by activating the NLRP3 inflammasome has not yet been fully elucidated. This study researched whether PM2.5 induces the NLRP3 inflammasomes activation to trigger pulmonary inflammation. Mice and MH-S cells were exposed to PM2.5, BOX5, and Rapamycin. Hematoxylin and eosin staining was performed on the lung tissues of mice. M1 macrophage marker CD80 expression in the lung tissues of mice and LC3B expression in MH-S cells was detected by immunofluorescence. IL-1β level in the lavage fluid and MH-S cells were detected by enzyme-linked immunosorbent assay. Protein expression was detected by Western blot. Autophagy assay in MH-S cells was performed by LC3B-GFP punctae experiment.PM2.5 exposure induced the lung injury of mice and increased NLRP3, P62, Wnt5a, LC3BII/I, and CD80 expression and IL-1β release in the lung tissues. PM2.5 treatment increased NLRP3, pro-caspase-1, cleaved caspase-1, Pro-IL-1β, Pro-IL-18, P62, LC3BII/I, and Wnt5a expression, IL-1β release, and LC3B-GFP punctae in MH-S cells. However, BOX5 treatment counteracted this effect of PM2.5 on lung tissues of mice and MH-S cells. Rapamycin reversed the effect of BOX5 on PM2.5-induced lung tissues of mice and MH-S cells.PM2.5 activated the NLRP3 inflammasome and IL-1β release in MH-S cells by facilitating the autophagy via activating Wnt5a. The findings of this study provided a new clue for the treatment of pulmonary inflammation caused by PM2.5.
Particulate matter 2.5 (PM2.5) is particles suspended in the atmosphere with a diameter of no more than 2.5 μm, which carries harmful substances such as heavy metals. PM2.5 has the characteristics of small size, large surface area, long propagation distance, and long stagnation time. It can be deposited in the alveoli after entering the body through the respiratory system. After penetrating the lung–blood barrier, PM2.5 can enter the blood circulatory system to cause damage to the tissues of multiple organs. Particularly, after being inhaled into the lung, PM2.5 can induce pulmonary inflammatory response through inducing the release of multiple inflammatory factors. Meanwhile, the harmful substances (such as polycyclic aromatic hydrocarbons) carried by PM2.5 can cause the production of free radicals in the lung tissues. Free radicals will break the balance of oxidants and antioxidants to induce the lung injury and pulmonary dysfunction. Therefore, the identification of the molecular mechanism in PM2.5-induced pulmonary inflammation is the key to treat this disease. Inflammasome is a main regulator of innate immunity, which participates in promoting the inflammatory response by activating caspase-1 and interleukin (IL)-1β. Nucleotide oligomerization domain (NOD)-like receptor protein 3 (NLRP3) inflammasome is an important part of body’s innate immunity. Pathogens, such as viruses, parasites, and bacteria, endoplasmic reticulum stress, and oxidative stress can trigger the activation of the NLRP3 inflammasome. As an intracellular multi-protein complex, NLRP3 inflammasome includes the expression of NLRP3, pro-caspase-1, and apoptotic speck protein. NLRP3 inflammasome can promote the release of inflammatory factors (such as IL-1β) by cleaving the pro-caspase-1 into caspase-1 via formatting an inflammasome complex. Meanwhile, NLRP3 inflammasome can activate the downstream complex signaling pathways to trigger a series of inflammatory reactions. As a pro-inflammatory cytokine, IL-1β is considered as an effector of the NLRP3 inflammasome, which participates in the process of the PM2.5-induced pulmonary inflammation. Previous study has been reported that PM2.5 could induce the pulmonary inflammation through activating the NLRP3. In the PM2.5-induced mouse lung injury model, PM2.5 has been found to activate the NLRP3 inflammasome to induce the over-production of IL-1β. However, the intrinsic molecular mechanism of PM2.5 activating NLRP3 has not yet been fully elucidated. Thus, this study was designed to research the intrinsic molecular mechanism of PM2.5 on the NLRP3 activation. Autophagy is an important metabolic process, which maintains the self-stable state of cells by removing the damaged proteins and organelles. Autophagy is involved in a variety of stress responses, such as inflammatory reaction. However, the over-activated autophagy can cause the cell damage and death, thereby triggering a series of inflammatory reactions. It has been revealed that PM2.5 could intensify the autophagy activation and induce the subsequent death of human bronchial epithelium cells. Intriguingly, the inhibition of the autophagy could protect against lung inflammation by suppressing the activation of the NLRP3 inflammasome. However, the enhanced autophagy could activate the NLRP3 inflammasome to augment the lung inflammation. Therefore, this study speculated that PM2.5 might induce the activation of the NLRP3 inflammasome via aggravating the autophagy. Recently, it has been found that PM2.5 induced the expression of inflammatory cytokines in human bronchial epithelium cells via activating the Wnt5a. Wnt5a played an important role in the autophagy of macrophages. However, whether PM2.5 activates the autophagy via inducing Wnt5a activation has not yet been elucidated. In our preliminary study, we used PM2.5 to treat B lymphocytes (WEHI-231 cells), T lymphocytes (EL4 cells), dendritic cells (JAWSII cells), and alveolar macrophage cells (MH-S cells) derived from lung tissues of mouse and then detect the expression of NLRP3 in the four cell lines. It was found that MH-S cell expressed the highest NLRP3 mRNA than the other three cell lines (as shown in Supplement Figure S1). Thus, this study used PM2.5 to treat C57BL/6 nude mice and MH-S cells. The aim was to verify the following speculation: PM2.5 might promote macrophages autophagy by activating Wnt5a, and then further promote the activation of the NLRP3 inflammasome to induce the pulmonary inflammation. The findings of this paper might provide a novel clue for the treatment of PM2.5-induced pulmonary inflammation.
The PM2.5 sample was collected from the building roof of the Second Hospital of Hebei Medical University. The PM2.5 sampler (air particulate sampler) was placed on the roof of the building from January 3, 2021 to February 3, 2021. The PM2.5 sampler was about 500 m away from the main traffic road with the height of about 35 m. The flow rate of the air particulate sampler was 100 L/min. The sources of PM2.5 were automobile exhaust, soil dust, industrial emissions, etc. That were emitted into the environment. The glass fiber filter paper was utilized for the collection of PM2.5. The PM2.5 was put into a freeze-drying bottle and placed in a refrigerator at −80°C. After being frozen and evacuated, the PM2.5 particles were dried into dry powder. After being sterilized, the PM2.5 dry powder was stored at −20°C.
PM2.5 sample was adhered to the double-sided carbon tape, and then fixed on the sample stage. After being gold-coated by a sputter coater (Desk V, Denton Vaccum), the morphology of PM2.5 sample was observed under a SEM (JSM-840, Jeol, Tokyo, Japan). EDS (JED-2300, Jeol, Tokyo, Japan) was applied for the analysis of the main elements in PM2.5. With a heated tungsten filament to be electron source, the images were obtained from PM2.5 sample. The acceleration voltage was 10 kV.
All protocols of animal handling and sampling were approved by the Institutional Animal Care and Use Committee of China Medical University (No. CMU2021111). All efforts were made to minimize the suffering of animals according to recommendations proposed by the European Commission (1997). The study was carried out in accordance with the approved protocol. All methods were conducted in accordance with relevant guidelines. C57BL/6 nude mice (n = 20, 5 weeks old, average weight of [22 ± 3] g) were commercially provided by Beijing Vital River Laboratory Animal Technology Co., Ltd (Beijing, China). All mice were housed in a 12 h day/night cycle room (at about 22°C) with free access to food and water. These mice were randomly divided into four groups: control group (n = 5), PM2.5 group (n = 5), PM2.5 + BOX5 group (n = 5), and PM2.5 + BOX5 + Rapamycin group (n = 5). Before using, the PM2.5 dry powder was dispersed into the sterilized phosphate buffered saline (PBS) to a concentration of 1 mg/mL. For mice of the control group, 10 μL of PBS was dripped into the nostrils. Mice of the PM2.5 group were nasally dripped with 10 μL of the PM2.5 PBS suspension (1 mg/mL). The nasally dripping was performed for 14 consecutive days twice daily (at 8:00 a.m. and 15:00 p.m.). For PM2.5 + BOX5 group, BOX5 (0.5 μg/mL, 10 μL) was nasally instilled into mice for two weeks with twice a week. PM2.5 PBS suspension (1 mg/mL, 10 μL) was then nasally dripped into mice for 14 consecutive days twice daily (at 8:00 a.m. and 15:00 p.m.). For PM2.5 + BOX5 + Rapamycin group, mice were nasally instilled with BOX5 (0.5 μg/mL, 10 μL) for two weeks with twice a week and administered rapamycin (4 mg/kg, dissolved in 0.25% Tween) for 6 consecutive days per week (for 2 weeks) through intraperitoneal injection. PM2.5 PBS suspension (1 mg/mL, 10 μL) was then nasally dripped into mice for 14 consecutive days twice daily (at 8:00 a.m. and 15:00 p.m.). On the 15 day, mice of each group were euthanized. The method of euthanizing the mice was as follows: Pentobarbital at a dose of 60 mg/kg (P0225, EKEAR Bio, Shanghai, China) was employed to deeply anesthetize the mice through intraperitoneal injection. Mice were considered to be deeply anesthetized if they had no response to head and limb stimulation. Then the mice were sacrificed via rapid cervical dislocation. The trachea and lungs of mice were separated and the right lung was ligated. The upper end of the trachea was inserted with a puncture needle and then ligated. The left lung was washed three times with PBS. The volume of PBS used per time was 0.3 mL. The lavage fluid was collected and centrifuged for 10 min at 1500 r/min/min and 4°C. After centrifugation, the supernatant of the lavage fluid was obtained and stored at −20°C for the following cytokine testing. The lung tissues of all mice were then collected and stored in a refrigerator at −80°C.
The right lung tissues of mice were fixed for 48 h in 4% paraformaldehyde. Then tissues were dehydrated by an ascending gradient of ethanol, followed by being embedded into paraffin. The lung tissues were cut into sections by using a rotary microtome to a thickness of 5 μm. Hematoxylin staining solution and eosin staining solution was sequentially utilized to stain the sections. The procedure of staining was performed in line with the directions. An ascending gradient of ethanol was used to dehydrate the sections. After treated by xylene, the sections were sealed in neutral gum and observed under a light microscope (Olympus, Japan).
Mouse alveolar macrophage cell line (MH-S) was purchased from American Type Culture Collection (ATCC, Manassas, VA, USA). RPMI-1640 medium containing 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 μg/mL streptomycin was used to culture MH-S cells at 37°C, 5% CO2.
The effect of PM2.5 on MH-S cell viability was explored by CCK-8 assay. MH-S cells were seeded into 96-well plates and cultured by RPMI-1640 medium containing 10% FBS and different concentration of PM2.5 (12.5, 25, 50, 100 and 200 μg/mL). After being cultured for 48 h at 37°C, 5% CO2, MH-S cells were incubated for 2 h with 10 μL of CCK-8 (Solarbio, Beijing, China) solution. The optical density value of each well was then measured using a porous microplate reader (BioTek, Winooski, VT USA). Five repeated wells were set for MH-S cells of each group. MH-S cells culture by RPMI-1640 medium containing 10% FBS were used as the control. The viability of MH-S cells was determined by the formula of (OD value of experimental group/OD value of control) × 100%.
By ultrasound, the sterilized PM2.5 dry powder was uniformly dispersed into RPMI-1640 medium (10% FBS) to a final working concentration of 100 μg/mL. MH-S cells (1 × 106 cells) were cultured in RPMI-1640 medium (1 mL) containing 10% FBS and 100 μg/mL PM2.5 for 8 h at 37°C, 5% CO2 (used as PM2.5 group). MH-S cells (1 × 106 cells) cultured in 1 mL of RPMI-1640 medium containing 10% FBS for 8 h were set as the control group. BOX5 was a specific inhibitor of Wnt5a. RPMI-1640 medium containing 10% FBS and BOX5 (100 μM) was used to pre-treat MH-S cells for 1 h at 37°C, 5% CO2. Then MH-S cells were cultured for 8 h with RPMI-1640 medium containing 10% FBS and 100 μg/mL of PM2.5 in the presence of BOX5. These cells were named the PM2.5 + BOX5 group. MH-S cells were pre-treated by RPMI-1640 medium containing 10% FBS, BOX5 (100 μM) for 1 h, and then treated by RPMI-1640 medium containing 10% FBS and Rapamycin (an autophagy activator, 3 μM) for another 1 h. Followed by this, these MH-S cells were cultured with RPMI-1640 medium containing 10% FBS and PM2.5 (100 μg/mL), BOX5 (100 μM), and Rapamycin (3 μM) for 8 h (used as the PM2.5 + BOX5 + Rapamycin group).
The LC3B expression in MH-S cells was detected by immunofluorescence. MH-S cells of each group were fixed by 4% paraformaldehyde for 15 min, followed by being incubated with 0.1% Triton-X for 10 min. Then 1% bovine serum albumin (BSA) was used to block cells for 1 h. MH-S cells were then treated by rabbit anti-LC3B (AF5225, Beyotime, Shanghai, China) antibody for 12 h at 4°C. Alexa Fluor 594-conjugated goat anti-rabbit secondary antibody (ab150080, Abcam, Shanghai, China) was added to incubate cells for 2 h at room temperature. The nuclei were stained by 4′, 6-diamidino-2-phenylindole (DAPI). LC3B expression (red fluorescence) were observed under a fluorescence microscope (CKX41, Olympus, Tokyo, Japan). The detection of M1 macrophage marker CD80 expression in mice lung tissues was implemented by immunofluorescence. The right lung tissues of mice were fixed for 48 h by 4% paraformaldehyde, and then dehydrated by ascending gradient ethanol. After being embedded into paraffin, the lung tissues were cut into sections (5 μm). The sections were sequentially treated by 0.1% Triton-X for 10 min, 1% BSA for 1 h, and then rabbit anti-CD80 primary antibody (ab215166, Abcam, Shanghai, China) overnight at 4°C. Alexa Fluor 488-conjugated goat anti-rabbit secondary antibody (ab150077, Abcam, Shanghai, China) was added onto the sections for 2 h treatment at room temperature. DAPI was dropped onto the sections for the nuclei staining. The expression of CD80 was observed under a fluorescence microscope (CKX41, Olympus, Tokyo, Japan).
MH-S cells of each group were collected and incubated by lysis buffer for 30 min on ice. The cell lysate were centrifuged for 15 min at 12,000 r/min/min and 4°C. The supernatant was collected to detect the level of IL-1β by ELISA. Additionally, the supernatant of mice lavage fluid was collected to detect the level of IL-1β by ELISA. The procedure was carried out by using an ELISA kit (CME0015-048, 4A Biotech, Beijing, China) according to the instructions.
MH-S cells of each group were cultured in the 6-well plates (1 × 105 cells per well). The LC3B-GFP virus (10 μL, BacMam 2.0, Thermo Fisher, Shanghai, China) was then added into each well to treat MH-S cells for 24 h. This process was carried out in line with the manufacturer’s protocol. After that, the LC3-GFP positive autophagasome were observed under a fluorescence microscope (CKX41, Olympus, Tokyo, Japan).
TEM was utilized to observe the ingestion of PM2.5 into MH-S cells. Briefly, MH-S cells of the control group, the PM2.5 group, the PM2.5 + BOX5 group, and the PM2.5 + BOX5 + Rapamycin group were cultured in the 6-well plates for 8 h with the relevant medium. Then cells were digested with trypsin and washed 3 times by PBS. Cells were fixed by 2.5% glutaraldehyde for 15 min and washed by PBS. Osmium tetroxide was used to fix cells for 90 min. Cells were then dehydrated by using an ascending gradient alcohol and ascending gradient acetone. Thereafter, cells were sealed in neutral resin and cut into slices with a thickness of 70–90 nm. Uranyl acetate and lead citrate were used to stain the sections according to the instructions. The sections were loaded onto copper grid and observed under a TEM (H-7650, Hitachi, Japan).
Lung tissues of mice and MH-S cells were treated with RIPA buffer (Solarbio, Beijing, China) for 30 min on ice to extract the total proteins. Protease Inhibitor Cocktail (Solarbio, Beijing, China) was pre-added into the RIPA buffer. The lysate samples were centrifuged at 14,000g and 4°C for 15 min to collect the supernatant. The total proteins in the supernatant underwent the sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) for the separation. After being transferred onto the polyvinylidene difluoride (PVDF) membranes, the proteins were blocked by 5% skimmed milk for 1 h at room temperature. Afterwards, primary antibodies were used to probe the proteins for 12 h at 4°C. The primary antibodies were rabbit anti-NLRP3 (1:1000, 15101, Cell Signaling Technology, Boston, USA), rabbit anti-P62 (1:1000, 18420-1-AP, Proteintech, Wuhan, China), rabbit anti-LC3B (1:1000, 2775, Cell Signaling Technology, Boston, USA), rabbit anti-Wnt5a (1:1000, MAB645, R&D, Minneapolis, USA), rabbit anti-pro-Caspase-1 (1:1000, ab179515, Abcam, Cambridge, UK), rabbit anti-cleaved Caspase-1 (1:1000, Cell Signaling Technology, Boston, USA), rabbit anti-pro-IL-1β (1:1000, WL02257, Wanlei Biotechnology, Shenyang, China), rabbit anti-pro-IL-18 (1:1000, FNab06765, Fine Biotechnology, Wuhan, China), and rabbit anti-β-actin (1:1000, ab8227, Abcam, Cambridge, UK). After washing twice using tris-buffered saline with 0.5% Tween-20 (TBST), the proteins were incubated with horseradish peroxidase (HRP)-labeled anti-rabbit secondary antibody (1:10,000, AS014, Abclonal, Wuhan, China) for 2 h at room temperature. Enhanced chemiluminescence (ECL) reagent (Solarbio, Beijing, China) was used to develop the proteins blots according to the instructions. The gray value of protein blots was evaluated by using an ECL Western blot detection kit (Amersham, Little Chalfont, UK). The operation was carried out strictly in line with the instructions. The blots were cut prior to hybridization with antibodies during blotting.
All experiments were independently repeated three times. Data was presented in the form of mean ± standard deviation. SPSS 19.0 software (SPSS, Inc. Chicago, IL, USA) was used for the data analysis. Two tailed paired Student’s t-test and one-way variance analysis (followed by Tukey’s post hoc test) were used, respectively, for the difference comparison between two groups and in more than two groups. p < .05 meant a statistically significant difference.
The effect of PM2.5 on the lung injury of mice was researched. H&E staining showed that, relative to the control group, lung tissues of mice in the PM2.5 group were severely damaged, as exhibited by the disordered alveolar structure, the thickened alveolar septal and monocyte-macrophage infiltration (Figure 1(a)). Western blot showed that, in comparison to the control group, significantly higher expression of NLRP3, P62, Wnt5a, and LC3BII/I proteins was found in mice lung tissues of the PM2.5 group (p < .05) (Figure 1(b)). ELISA exhibited markedly higher IL-1β level in mice lavage fluid of the PM2.5 group than the control group (p < .05) (Figure 1(c)). The expression of M1 macrophage marker CD80 in mice lung tissues was detected by immunofluorescence to evaluate the M1 polarization of macrophages. Relative to the control group, mice of the PM2.5 group showed the intensified staining of CD80, indicating the increased M1 macrophages in lung tissues of the PM2.5 treated mice (Figure 1(d)). Therefore, PM2.5 treatment induced the lung injury of mice and activated the NLRP3 inflammasome, Wnt5a, autophagy, and IL-1β release in mice lung tissues.
This study used BOX5 (a specific inhibitor of Wnt5a) and Rapamycin (an autophagy activator) to treat the PM2.5-induced mice. The lung tissue injury of mice was detected by H&E staining. As shown in Figure 2(a), no obviously damage was observed in the lung tissues of the control group mice. However, mice of the PM2.5 group showed the severally lung injury, such as the disorganized alveolar architecture, the thickened alveolar septa, and the markedly infiltrated mononuclear macrophages. Relative to the PM2.5 group, mice of the PM2.5 + BOX5 group exhibited much attenuated lung injury, as manifested by the decreased alveolar structural disorder, the decreased alveolar septal thickening and the decreased monocyte-macrophage infiltration. However, Rapamycin treatment reversed the effect of BOX5 on the lung injury in the PM2.5-induced mice. Additionally, Western blot was utilized to monitor the expression of NLRP3, P62, LC3BI, LC3BII, and Wnt5a proteins in lung tissues (shown in Figure 2(b)). Mice of the PM2.5 group expressed distinctly higher proteins of Wnt5a, NLRP3, P62, and LC3BII/I in lung tissues than the control group (p < .05). Intriguingly, mice of the PM2.5 + BOX5 group expressed lower NLRP3, P62, Wnt5a, and LC3BII/I proteins in lung tissues than the PM2.5 group (p < .05). However, in comparison to the PM2.5 + BOX5 group, much higher NLRP3, P62, and LC3BII/I proteins was observed in the lung tissues of the PM2.5 + BOX5 + Rapamycin group mice (p < .05). The expression of Wnt5a protein was much lower in the lung tissues of the PM2.5 + BOX5 + Rapamycin group mice when relative to the PM2.5 group (p < .05). The expression difference of Wnt5a protein in mice lung tissues was not obvious between the PM2.5 + BOX5 group and the PM2.5 + BOX5 + Rapamycin group. The IL-1β level in mice lavage fluid was explored by ELISA, and the results are shown in Figure 2(c). Mice of the PM2.5 group had higher IL-1β level in lung tissues than the control group (p < .05). Conversely, lower IL-1β level was observed in mice lavage fluid of the PM2.5 + BOX5 group when compared to the PM2.5 group (p < .05). Intriguingly, relative to the PM2.5 + BOX5 group, higher IL-1β level was occurred in mice lavage fluid of the PM2.5 + BOX5 + Rapamycin group (p < .05). According to immunofluorescence, the intensified CD80 staining was occurred in the lung tissues of the PM2.5 group mice when compared to the control group. Oppositely, the attenuated CD80 staining was found in mice lung tissues of the PM2.5 + BOX5 group when compared to the PM2.5 group. Matched to the PM2.5 + BOX5 group, mice of the PM2.5 + BOX5 + Rapamycin group displayed the enhanced CD80 staining in the lung tissues (Figure 2(d)). Thus, all of these data implied that the inhibition of Wnt5a attenuated the PM2.5 induced lung injury in mice and the activation of autophagy counteracted this effect.
The morphology of PM2.5 was observed using the SEM. The size of PM2.5 was no more than 2.5 μm in the diameter. The morphology of PM2.5 was irregular, which exhibited as rod-like and granular shapes (Figure 3(a)). EDS was used for the elemental analysis of PM2.5. The element map of PM2.5 showed that, the elements included in PM2.5 contained carbon (C), oxygen (O), sodium (Na), magnesium (Mg), aluminum (Al), silicon (Si), Aurum (Au), sulfur (S), potassium (K), calcium (Ca), and barium (Ba) (Figure 3(b)). The effect of PM2.5 on the viability of MH-S cells was explored by CCK-8 assay. PM2.5 at a concentration of 12.5, 25, 50, and 100 μg/mL could not obviously affect the viability of MH-S cells. However, PM2.5 at a concentration of 200 μg/mL much reduced the viability of MH-S cells (Figure 3(c)). Thus, this study used 100 μg/mL of PM2.5 to treat MH-S cells in the following study.
As shown in Figure 4(a), prominently higher expression of NLRP3, pro-caspase-1, cleaved caspase-1, pro-IL-1β, and pro-IL-18 proteins were discovered in MH-S cells of the PM2.5 group when relative to the control group (p < .05). In comparison to the control group, the IL-1β level in MH-S cells of the PM2.5 group was much increased (p < .05) (Figure 4(b)). These data indicated that PM2.5 treatment activated the NLRP3 inflammasome and exacerbated the IL-1β release in MH-S cells.
Western blot displayed a significantly higher expression of P62 and LC3BII/I proteins in MH-S cells of the PM2.5 group when compared with the control group (p < .05) (Figure 5(a)). Simultaneously, the expression of Wnt5a protein in MH-S cells of the PM2.5 group was higher than the control group (p < .05) (Figure 5(b)). Immunofluorescence exhibited that, the LC3B expression (red fluorescence) was higher in MH-S cells of the PM2.5 group than the control group (Figure 5(c)). From Figure 5(d), more LC3B-GFP positive autophagasome (green fluorescence) were found in MH-S cells of the PM2.5 group when matched to the control group. Thus, the autophagy and expression of Wnt5a protein in MH-S cells was activated by the PM2.5 treatment.
BOX5 (a specific inhibitor of Wnt5a) and Rapamycin (an autophagy activator) were used to treat MH-S cells. As shown in Figure 6(a), higher expression of Wnt5a, NLRP3, pro-caspase-1, cleaved caspase-1, pro-IL-1β, and pro-IL-18 proteins were observed in MH-S cells of the PM2.5 group, in comparison to the control group (p < .05). Interestingly, much lower expression of Wnt5a protein was found in MH-S cells of the PM2.5 + BOX5 group and the PM2.5 + BOX5 + Rapamycin group when matched to the PM2.5 group (p < .05). MH-S cells of the PM2.5 + BOX5 group had markedly lower expression of NLRP3, pro-caspase-1, cleaved caspase-1, pro-IL-1β, and pro-IL-18 proteins than the PM2.5 group (p < .05). However, in comparison to the PM2.5 + BOX5 group, the expression of NLRP3, pro-caspase-1, cleaved caspase-1, pro-IL-1β, and pro-IL-18 proteins were all significantly increased in MH-S cells of the PM2.5 + BOX5 + Rapamycin group (p < .05). ELISA presented remarkably higher IL-1β level in MH-S cells of the PM2.5 group when relative to the control group (p < .05). Oppositely, lower IL-1β level in MH-S cells of the PM2.5 + BOX5 group was found when compared with the PM2.5 group (p < .05). Relative to the PM2.5 + BOX5 group, the increased IL-1β level was observed in MH-S cells of the PM2.5 + BOX5 + Rapamycin group (p < .05) (Figure 6(b)). Hence, PM2.5 might activate the NLRP3 inflammasome and the IL-1β release in MH-S cells by facilitating the autophagy via activating Wnt5a.
According to Western blot, the expression of Wnt5a, P62, and LC3BII/I proteins was significantly increased in MH-S cells of the PM2.5 group when matched to the control group (p < .05). Lower Wnt5a protein was exhibited in MH-S cells of the PM2.5 + BOX5 group and the PM2.5 + BOX5 + Rapamycin group when compared to the PM2.5 group (p < .05). Intriguingly, remarkably lower expression of P62 and LC3BII/I proteins was observed in MH-S cells of the PM2.5 + BOX5 group when matched to the PM2.5 group (p < .05). Oppositely, the expression of P62 and LC3BII/I proteins was significantly elevated in MH-S cells of the PM2.5 + BOX5 + Rapamycin group when relative to the PM2.5 + BOX5 group (p < .05) (Figure 7(a)). LC3B expression in MH-S cells was detected using immunofluorescence. The enhanced LC3B staining (red fluorescence) was shown in MH-S cells of the PM2.5 group relative to the control group. Intriguingly, LC3B staining was attenuated in MH-S cells of the PM2.5 + BOX5 group than the PM2.5 group. However, matched to the PM2.5 + BOX5 group, MH-S cells of the PM2.5 + BOX5 + Rapamycin group exhibited the intensified LC3B staining (Figure 7(b)). Moreover, in comparison to the control group, MH-S cells of the PM2.5 group presented the enhanced LC3B-GFP positive autophagasome. However, less LC3B-GFP positive autophagasome were observed in MH-S cells of the PM2.5 + BOX5 group when relative to the PM2.5 group and the PM2.5 + BOX5 + Rapamycin group (Figure 7(c)). These results revealed that PM2.5 facilitated the autophagy in MH-S cells via activating Wnt5a.
TEM was utilized to observe the ingestion of PM2.5 into MH-S cells. The images are shown in Figure 8. The red arrow represented autophagosomes, and the blue arrow represented PM2.5. MH-S cells of the control group had less autophagosomes than the PM2.5 group. Some PM2.5 particles were taken into MH-S cells of the PM2.5 group, the PM2.5 + BOX5 group and the PM2.5 + BOX5 + Rapamycin group. Less autophagosomes were observed in MH-S cells of the PM2.5 + BOX5 group when relative to the PM2.5 group. Matched to the PM2.5 + BOX5 group, MH-S cells of the PM2.5 + BOX5 + Rapamycin group had more autophagosomes. It is well known that the increased autophagosomes meant the enhancement of autophagy. Therefore, PM2.5 increased the autophagy in MH-S cells. However, Wnt5a inhibitor (BOX5) attenuated the PM2.5-induced autophagy in MH-S cells. Intriguingly, Rapamycin abrogated the effect of BOX5 on the PM2.5-induced autophagy in MH-S cells.
In recent decades, severe air pollution has posed a serious threat to the health of human, especially PM2.5 in the polluted air. In our previous study, PM2.5 has been detected to carry a large amount of harmful components, such as polycyclic aromatic hydrocarbons. This study detected that PM2.5 contained several kinds of harmful elements, such as Au, Si, and S. The onset and development of many respiratory diseases has been reported to be related to PM2.5 in the atmosphere. In this study, PM2.5 exposure induced mice lung injury and activated the NLRP3 inflammasome, Wnt5a, and IL-1β release in mice lung tissues and MH-S cells. Interestingly, BOX5 treatment counteracted this effect of PM2.5. Moreover, Rapamycin abrogated the effect of BOX5 on the PM2.5-induced damage on mice lung tissues and MH-S cells. Thus, PM2.5 might activate the NLRP3 inflammasome and IL-1β release in MH-S cells by facilitating the autophagy via activating Wnt5a. The mechanism diagram is shown in Figure 9 (created by the authors). The innate immune system can respond initially to the irritants, and innate immune cells can detect pathogens with a fixed number of germline-encoded pattern recognition receptors. Pattern recognition receptors detect the unique microbial structure called “pathogen-associated molecular pattern (PAMP)” Interestingly, the damaged cells can trigger the pattern recognition receptors through releasing the danger-associated molecular pattern (DAMP). After experiencing PAMP and DAMP, pattern recognition receptors can trigger the signaling cascades and then activate the nuclear factor-κB (NF-κB) to facilitate the formation of an inflammasome. Pattern recognition receptor is a main component of inflammasome. It can bind the nucleotide-binding oligomerization domain-like receptors (NLRs) to effector pro-caspase-1, thereby activating caspase-1. The activated caspase-1 further cleaves pro-IL-1β into IL-1β to induce the pro-inflammatory response. It has been identified that PM2.5 could activate the inflammasome through activating the NF-κB signaling. The activation of the NLRP3 inflammasome is associated with the lung inflammation. PM2.5 can promote the inflammatory response by activating the NLRP3 inflammasome. Recently, the activation of the NLRP3 inflammasome has been found to be involved in the pulmonary inflammation induced by PM2.53. More data about PM2.5 regulating the activation of NLRP3 to trigger lung inflammation are rarely to be found. In this study, PM2.5 exposure enhanced the expression of NLRP3, pro-caspase-1 and cleaved caspase-1 proteins and the release of IL-1β in mice lung tissues and MH-S cells. Typically, caspase-1 can be activated within the NLRP3 inflammasome, which further lead to the cleavage of pro-IL-1β into mature IL-1β. Thereafter, IL-1β releases to cells to cause the inflammation of cells and tissues. NLRP3 and pro-caspase-1 proteins are two important markers of the activated NLRP3 inflammasome. Pro-IL-18 is one of the major components of the NLRP3 inflammasome, which can cause the inflammatory damage to lung tissues. This paper indicated that PM2.5 treatment induced the expression of pro-IL-18 in mice lung tissues. Moreover, it is well known that M1 macrophages exert a pro-inflammatory function through releasing the pro-inflammatory factors (such as IL-1β) and then exacerbate the inflammatory responses in lung tissues. In this paper, PM2.5 increased the expression of M1 macrophage marker CD80 in mice lung tissues, indicating the M1 polarization of macrophages triggered by the PM2.5 treatment. This study confirmed that PM2.5 induced the inflammation in mice lung tissues and MH-S cells by activating the NLRP3 inflammasome. Autophagy is ubiquitous in eukaryotic cells, which is a highly conserved and homeostatic process. Under normal circumstances, autophagy can maintain the stability of the intracellular environment by removing the damaged organelles, etc. However, over activation of autophagy can induce the excessive degradation of cellular contents, thereby resulting in the impaired cell function and autophagic death. Previous research has reported that autophagy was over-induced in the damaged lung tissues caused by PM2.5. P62 is a classical receptor of autophagy. In mouse model exposed to PM2.5, PM2.5 has been revealed to cause the lung tissue damage by triggering autophagy via increasing the expression of P62. PM2.5 triggered the reproductive toxicity to male rats by inducing autophagy activation via promoting the expression of P62. Similarly, this study indicated that PM2.5 exposure enhanced the expression of P62 protein in mice lung tissues and MH-S cells. LC3B is the most important subtype of LC3, which is an important marker of autophagy. The elevated LC3BII/I ratio has become the main criterion for judging autophagy level. A previous study has reported that PM2.5 treatment induced the activation of autophagy in A549 cells by increasing the expression of LC3BII. In this paper, PM2.5 treatment promoted the expression of LC3B protein in MH-S cells and elevated the LC3BII/I ratio in mice lung tissues and MH-S cells. These data indicated that PM2.5 treatment activated the autophagy in mice lung tissues and MH-S cells. Previous study has reported that the NLRP3 inflammasome could be activated by autophagy. Similarly, this study revealed that, PM2.5 could activate the NLRP3 inflammasome and IL-1β release in mice lung tissues and MH-S cells by facilitating the autophagy. It has been found that PM2.5 increased the expression of Wnt5a in human bronchial epithelial cells. However, the activation of Wnt5a could facilitate the inflammatory response in lung tissues. To data, whether PM2.5 induces lung injury by activating Wnt5a has not been fully elucidated. This paper revealed that PM2.5 increased the expression of Wnt5a in mice lung tissues and MH-S cells. BOX5 (a specific antagonist of Wnt5a) treatment could reverse the promoting effect of PM2.5 on the lung injury, the NLRP3 inflammasome activation, IL-1β release, and autophagy in mice or MH-S cells. Thus, PM2.5 might activate the activation of the NLRP3 inflammasome and promoted the IL-1β release and autophagy in mice lung tissues and MH-S cells by activating Wnt5a. Moreover, Rapamycin (an autophagy activator) treatment reversed the inhibitory effect of BOX5 on the NLRP3 inflammasome activation, IL-1β release and autophagy in mice lung tissues and MH-S cells induced by PM2.5. Therefore, PM2.5 might activate the NLRP3 inflammasome and IL-1β release in MH-S cells by facilitating the autophagy via activating Wnt5a. Of course, there were limitations in this study. First, macrophage recruitment or clearance in lung tissues upon PM2.5 treatment, PM2.5 + BOX treatment and PM2.5 + BOX + Rapamycin treatment should be investigated. Second, from Figure 1(a), (h), (e) staining indicated that PM2.5 treated mice exhibited more collagen components in the lung tissues. This suggests that PM2.5 treated mice might have lung fibrosis. Masson staining should be performed to verify this speculation. Moreover, a co-staining of macrophage maker and target proteins in mice lung tissues should be performed. Finally, sample size calculation should be done in this study. However, due to the laboratory limitations, these experiments cannot be performed at present. This issue will be focus of our future research.
PM2.5 treatment induced mice lung injury, increased the expression of Wnt5a, and activated the NLRP3 inflammasome, IL-1β release, and autophagy in mice lung tissues and MH-S cells. The inhibition of Wnt5a attenuated the lung injury of the PM2.5-induced mice. More importantly, PM2.5 might activate the NLRP3 inflammasome and IL-1β release in MH-S cells by facilitating the autophagy via activating Wnt5a. This paper provided a novel clue for pulmonary inflammation treatment induced by PM2.5. Clinically, pulmonary inflammation induced by PM2.5 could be treated by targeting the inhibition of Wnt5a. Of course, more research still needs to be done in the future.
Click here for additional data file. Supplemental Material for PM2.5 activated NLRP3 inflammasome and IL-1β release in MH-S cells by facilitating autophagy via activating Wnt5aGuanli Yuan, Yinfeng Liu, Zheng Wang, Xiaotong Wang, Zhuoxiao Han, Xixin Yan and Aihong Meng in International Journal of Immunopathology and Pharmacologyr |
PMC9647319 | 34269117 | Valentina Malafoglia,Sara Ilari,Laura Vitiello,Michael Tenti,Eleonora Balzani,Carolina Muscoli,William Raffaeli,Antonello Bonci | The Interplay between Chronic Pain, Opioids, and the Immune System | 16-07-2021 | pain,chronic pain,chronic pain biomarkers,chronic pain diagnosis,opioids,opioid receptors,immune system,immune cells,opioid receptor signalling pathway,neuroimmune synapse | Chronic pain represents one of the most serious worldwide medical problems, in terms of both social and economic costs, often causing severe and intractable physical and psychological suffering. The lack of biological markers for pain, which could assist in forming clearer diagnoses and prognoses, makes chronic pain therapy particularly arduous and sometimes harmful. Opioids are used worldwide to treat chronic pain conditions, but there is still an ambiguous and inadequate understanding about their therapeutic use, mostly because of their dual effect in acutely reducing pain and inducing, at the same time, tolerance, dependence, and a risk for opioid use disorder. In addition, clinical studies suggest that opioid treatment can be associated with a high risk of immune suppression and the development of inflammatory events, worsening the chronic pain status itself. While opioid peptides and receptors are expressed in both central and peripheral nervous cells, immune cells, and tissues, the role of opioids and their receptors, when and why they are activated endogenously and what their exact role is in chronic pain pathways is still poorly understood. Thus, in this review we aim to highlight the interplay between pain and immune system, focusing on opioids and their receptors. | The Interplay between Chronic Pain, Opioids, and the Immune System
Chronic pain represents one of the most serious worldwide medical problems, in terms of both social and economic costs, often causing severe and intractable physical and psychological suffering. The lack of biological markers for pain, which could assist in forming clearer diagnoses and prognoses, makes chronic pain therapy particularly arduous and sometimes harmful. Opioids are used worldwide to treat chronic pain conditions, but there is still an ambiguous and inadequate understanding about their therapeutic use, mostly because of their dual effect in acutely reducing pain and inducing, at the same time, tolerance, dependence, and a risk for opioid use disorder. In addition, clinical studies suggest that opioid treatment can be associated with a high risk of immune suppression and the development of inflammatory events, worsening the chronic pain status itself. While opioid peptides and receptors are expressed in both central and peripheral nervous cells, immune cells, and tissues, the role of opioids and their receptors, when and why they are activated endogenously and what their exact role is in chronic pain pathways is still poorly understood. Thus, in this review we aim to highlight the interplay between pain and immune system, focusing on opioids and their receptors.
Chronic pain (CP) places an enormous burden on sufferers and health care systems, leading the European Commission to include pain as a topic in its mission-oriented research (https://ec.europa.eu/health/sites/health/files/policies/docs/ev_20181112_co06_en.pdf; Thematic Network on the Societal Impact of Pain—Recommendations for Policy Actions—05 November 2018). CP is not only a complex health problem but also a real “bio-psycho-social experience,” which requires a more comprehensive approach than many other pathologies (Darnall and others 2017). Even if pain is considered more frequently as a specific symptom of another health issue, CP represents the main and exclusive condition for many patients (Raffaeli and Arnaudo 2017), as underlined in the new “International Classification of Diseases–11th Revision (ICD-11) (Treede and others 2019). Despite the large number of preclinical studies (Burma and others 2017), the intrinsic subjectivity of this pathology makes clinical diagnosis still quite arduous. In particular, pain intensity is the most difficult CP characteristic to quantify and is usually described through a patient self-certification based on numerical, visual, verbal, or facial pain ranking scales. In this context, the gap between preclinical and clinical CP data are larger than for other diseases. Moreover, the lack of specific CP biomarkers complicates diagnostic and prognostic approaches, often resulting in a delay of the appropriate pharmacological treatments (Borsook and others 2011). In CP therapy, opioids are commonly used, but their beneficial effects are often overshadowed by their side effects (Cumenal and others 2021). This is also due to the absence of proper, universally shared and agreed-upon therapeutic guidelines about the right dosages of opioid medication. For the reasons mentioned above, more and more scientists are highlighting the need to strengthen research aimed at discovering novel pain biomarkers. Quantifiable biomarkers of pain would facilitate the identification of new mechanisms and signs to help with pain diagnoses, predict subgroups inside heterogeneous painful conditions, and provide pharmacodynamic modifications associated with a specific dosage and drug formulation (Zhao and others 2015). In this context, recent studies have paid attention to promising peripheral pain biomarkers, which represent an efficient and highly feasible avenue for investigation, given that all is needed is a blood sample (Backryd 2015). For example, Niculescu and colleagues, through a blood microarray analysis (Niculescu and others 2019) have shown that several genes and, recently, specific microRNAs expressions (Tavares-Ferreira and others 2019), are tightly correlated with the intensity of pain. The interest in peripheral detection of pain biomarkers finds its origin in the hypothesis that hyperalgesia is a result of the interplay between the immune and nervous systems, where the bridge between the two is represented by opioid receptors, expressed in both the central and the peripheral nervous system (Campana and others 2010; Kipnis 2016). In this context, a recent study proposed that Mu opioid receptors (MOR) expressed on B cells may be a potential biological marker (Mu-Lympho-Marker, MLM) to assist health care professionals in crafting a more objective chronic pain diagnosis, both in patients suffering from fibromyalgia (FM) and osteoarthritis (OA) (Raffaeli and others 2020). The major goal of this review is to explore and discuss the latest findings in chronic pain biomarkers in the immune and nervous systems, in order to disentangle the relationship between opioid receptors, uncontrolled pain conditions, and the immune system.
For the past four decades, pain has been defined as “An unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage” (Treede 2018). However, the term “describe” has been improperly used, considering that some patients are unable to verbally express their painful experience. Thus, in 2020, the International Association for the Study of Pain (IASP) approved the new definition of pain as “An unpleasant sensory and emotional experience associated with, or resembling that associated with, actual or potential tissue damage” (Raja and others 2020), where the subjectivity of a single person is highlighted by both sensory and affective components. Under physiological conditions, a noxious stimulus activates first order neurons in order to reach specific laminae of the spinal cord. There, second order neurons send the information to different zones of the brain where it is processed. The intensity and the body localization of the stimulus are analyzed at the level of the somatosensory cortex. The cingulate and insular cortices, contribute to the affective component of pain, by receiving projection neurons, via connections in the parabrachial nucleus and amygdala. Then, the rostral ventral medulla and midbrain periaqueductal gray drive the descending feedback system, regulating the output from the spinal cord (Basbaum and others 2009) (Fig. 1). When this physiological pathway undergoes a pathological alteration and pain persists beyond the resolution of the underlying disorder, the healing of an injury, or sometimes without any cause, this is usually called “chronic pain.” Although CP is considered a persistent or recurrent pain lasting longer than 3 months (Treede and others 2015), it is not appropriate to call it a mere temporal extension of acute pain. In CP, the symptom represents the disease itself and this “bio-psycho-social experience” requires a multidisciplinary approach (Darnall and others 2017; Raffaeli and Arnaudo 2017). The main problem in CP remains its accurate diagnosis, particularly due to the lack of specific tools that detect subjective pain intensity. Today, evaluation of CP still consists of a patient self-certification using on various ranking scales (Thong and others 2018) (Fig. 2).
In the past decades, the need for CP biomarkers for a specific diagnosis, prognosis, and therapy, has grown exponentially. The concept that pain is a subjective conscious perception, which requires brain activity, has placed brain processes at the center of pain biomarker research (Reckziegel and others 2019). Functional magnetic resonance imaging (fMRI) (Kumbhare and others 2017), positron emission tomography (PET) (Albrecht and others 2019), and proton magnetic resonance spectroscopy (H-MRS) (Levins and others 2019) are increasingly used to detect structural, functional, and neurochemical information as potential biomarkers for CP. However, brain imaging is a complex field, and new findings and approaches are needed to bypass overlapping information due to chronic pain side effects and comorbid conditions (e.g., anxiety, depression, and fatigue) (Reckziegel and others 2019). Recently, Gunn and others (2020) presented an alternative retrospective biomarker assay, revealing an atypical biochemistry in CP patients. They proposed a panel of functional biomarkers that underlines the role of oxidative stress, cytokine-mediated inflammation, neurotransmitters, and micronutrients in CP conditions (Gunn and others 2020). In particular, the study showed elevated levels of quinolinic acid, pyroglutamate, xanthurenic acid, acrolein metabolite 3-hydroxypropyl mercapturic acid, and methylmalonic acid. They also detected abnormally low levels of 5-hydroxyindoleacetate (metabolite of serotonin) and vanilmandelate (metabolite of norepinephrine). However, more data in this context are needed, considering that medications and other conditions not associated with chronic pain were not analyzed in this study, and may be potential additional causes in the occurrence of atypical biomarkers. A key feature of a good biomarker is that it has to be easy and fast to detect (Califf 2018). This is why new studies are focusing on peripheral diagnostic approaches involving blood samples (Backryd 2015). For example, Niculescu and others (2019) have shown a blood microarray analysis can identify severe pain risk genes, which are more strongly expressed than putative protective/resilience genes, with the goal of developing a pain markers gene expression database. Other studies have demonstrated the correlation between specific microRNAs expressions (Tavares-Ferreira and others 2019) and CP, such as in low back pain (Hasvik and others 2019), osteoarthritis (Swingler and others 2019), and neuropathic pain (Tavares-Ferreira and others 2019). However, the lack of a miRNAs database for the healthy population and another one specific for CP patients makes miRNA validation as CP biomarkers quite difficult. Moreover, every disease ideally should have a specific miRNA of reference in order to bypass overlapping data (Lopez-Gonzalez and others 2017; Ramanathan and Ajit 2016). Recently, a new study has proposed a pain “molecular fingerprint” construction, using a vibrational spectroscopy technique, in order to identify metabolites associated with different CP conditions as serologic biomarkers (Hackshaw and others 2019). Together, these studies highlight the critical need and utility of blood samples in this field (Fig. 3).
As mentioned above, peripheral detection of pain biomarkers is based on the idea that hyperalgesia is a consequence of the interplay between immune and nervous systems, where the interface is represented by opioids receptors (Campana and others 2010; Kipnis 2016). Endogenous opioids derive from the precursors proopiomelanocortin, proenkephalin, and prodynorphin, which respectively encode for β-endorphin, enkephalins (Met-enkephalin and Leu-enkephalin), and dynorphins (Bodnar 2017). Opioid peptides share a common opioid-motif containing the Tyr-Gly-Gly-Phe-Met/leu amino acid sequence and explicate their action by binding to specific receptors. β-Endorphin and enkephalins are antinociceptive peptides, linking µ (mu [MOR]/Oprm1) and δ (delta [DOR]/Oprd1) opioid receptors (Hurley and Hammond 2001; Smith and others 1992). Dynorphins can mediate not only antinociceptive effects, via κ (kappa [KOR]/Oprk1) opioid receptors but also pro-nociceptive effects via N-methyl-d-aspartate (NMDA) receptors (Podvin and others 2016; Stein 2016). Opioid receptors are G-protein-coupled receptors, expressed in the central and peripheral nervous systems, and in neuroendocrine and immune tissues and cells (Stein 2016). Endogenous opioids and their receptors mediate analgesia, but also tolerance and dependence after long-term exposure to exogenous full opioid agonists, and withdrawal syndrome after an abrupt interruption of exogenous opioid exposure or after antagonist administration (Stein 2018). MOR mediates side effects such as nausea, sedation, respiratory depression, reward/euphoria, biliary spasms and constipation, urinary retention, and reduced gastrointestinal motility. KOR can lead to aversive, sedative, and diuretic effects, as well as dysphoria. DOR is involved in respiratory depression, constipation, and reward and euphoria. In the 1990s, a new endogenous opioid system was identified. The ligand was defined as nociceptin/orphanin (N/OFQ) and the receptor as nociceptin orphanin peptide receptor (NOP receptor), also called opioid receptor like1 (ORL1) receptor. The sequence of N/OFQ is closely related to that of dynorphin A. Furthermore, N/OFQ is not active at the classical opioid receptors, such as mu, kappa, and delta receptors. Unlike other members of the opioid family, N/OFQ plays a crucial role in pain modulation in a bidirectional manner, exhibiting either pro- or anti-nociceptive effects, depending on a series of complex factors, including type of pain, administration strategies and dosage of opioid agonists (Toll and others 2019). NOP is a G-coupled receptor (Agostini and Petrella 2014; Kiguchi and others 2016). This receptor is involved in spinal analgesia and mediates side effects such as tolerance to morphine after chronic treatment, hyperalgesia at low doses of N/OFQ, exacerbation of CP condition, depression, and Parkinson’s disease (Table 1).
Opioid receptors play a role in synaptic pain transmission, which takes place between C-fiber primary afferents and secondary projection neurons in the dorsal horn of the spinal cord. After binding their specific receptors, opioids lead to the dissociation of the trimeric G-protein complex, which switches from a GDP-bound inactive to a GTP-bound active state, into Gα and Gβγ subunits. Gβγ subunits can directly bind calcium (Ca++) channels and inhibit intracellular Ca++ influx. Activated Gα subunits can inhibit adenylyl cyclase (AC) activity and the consequential production of cyclic AMP (cAMP) and, in a downstream mechanism, protein kinase A (PKA) and Ca++ ion influx. In fact, Ca++ ions have a pivotal role in the synaptic process, enhancing glutamate (Glu) release from presynaptic vesicles. Glu mediates fast excitatory transmission between primary and secondary sensory neurons, by binding ionotropic α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and NMDA receptors, and slow transmission by acting on metabotropic glutamate receptors (mGluRs). Slow synaptic effects are also mediated by other neuropeptides, such as substance P and calcitonin gene-related peptide (CGRP), through metabotropic G protein-coupled receptors and receptor tyrosine kinases. Moreover, opioid peptides act both pre- and postsynaptically to inhibit the pain pathway. Presynaptic inhibition reduces voltage-gated calcium (Ca++) channel activity, while postsynaptic inhibition enhances chloride (Cl−) influx and potassium (K+) efflux (Golan 2008). These events prevent excitation and propagation of action potentials in second order projection neurons and suppresses pain development (Fig. 4). Peripherally, at the site of injury, the opioid receptor-mediated synaptic pain pathway involves peripheral nociceptors, leukocytes, and anti-inflammatory cytokines (Pinho-Ribeiro and others 2017). Peripheral opioid receptor-containing immune cells reach the inflamed tissue, and their activation leads to the secretion of opioid peptides, which bind their specific peripheral neuronal opioid receptors. At the early stage of the inflammation process, granulocytes are the major opioid-containing leukocytes involved, later monocytes, macrophages, and lymphocytes predominate (Brack and others 2004). Not all the totality of immune cells produce opioids, but it is well known that, during inflammation and leukocytes’ homing, the expression of opioid peptides increases (Hua 2016; Mousa and others 2007). Opioid release is triggered by several endogenous factors, including temperature, low pH, proteolytic activity, or local inflammatory factors (Julius and Basbaum 2001). When released, opioid peptides penetrate the perineural sheath and activate specific receptors on peripheral terminals of sensory neurons, producing analgesia and also anti-inflammatory effects such as the inhibition of substance P, NA (noradrenaline) and TNF (tumor necrosis factor)-α neuronal release (Mambretti and others 2016; Stein and Machelska 2011). This mechanism produces analgesia through the inhibition of activity of nociceptors (Rittner and others 2008; Stein and Kuchler 2012). Recently, in a mouse model, an expansion of the classical model of analgesia mediated by opioid receptors was presented, focusing on immune cells containing opioid receptors (Celik and others 2016; Machelska and Celik 2020). In contrast to the previously described model (Fig. 3), in this new model leukocytes opioid receptor mediate analgesia through the release of pain-inhibiting opioid peptides. Gi proteins are still involved in the process, in association with the phospholipase C (PLC), which leads to production of the second messenger inositol 1,4,5-trisphosphate (IP3). IP3 binds and activates the IP3 receptor (IP3R) in the endoplasmic reticulum (ER), leading to the release of intracellular Ca++ ions. This pathway permits extracellular delivery of intracellular Ca++-dependent opioid peptides, which bind specific opioid receptors on peripheral nerves membranes, thereby inhibiting pain (Fig. 5).
In a rat study, researchers showed that the half-life of endogenous peptides at the site of inflammation is strictly conditioned by peripheral blood proteases: 5 minutes for enkephalins and 40 minutes for β-endorphins. Thus, for the best analgesic effects, exogenous opioids peptides should be ideally released in close proximity to sensory neurons (Hua and others 2006). Along these lines, several studies provide evidence for the close association between peripheral nerve and opioid-containing immune cells (Hua 2016). Leukocytes have been found physically close to the innervation of many organs, such as the skin (Darsow and Ring 2001), eye (Schafer and others 1994), liver (Kaiser and others 2003), respiratory tract (Kingham and others 2002), and gastrointestinal tract (Tournier and Hellmann 2003). In fact, a bidirectional interaction between immune cells and primary afferent nerves has been described and is supported by three major observations. First, several studies indicate a direct membrane-membrane contact (Crivellato and others 2002; Tian and others 2000) when nerve fibers end on the surface of lymphoid organs. This anatomical connection chemically mediates the bidirectional release of transmitters and postsynaptic receptors activation (Stein 2013). Second, both immune cells and neurons share common ligands and receptors (Shaw and Allen 2001). Third, this ligand-receptor communication influences and activates cellular pathways in both immune and nervous systems (Rittner and others 2008). Traditionally, the term “synapse” indicates a stable adhesive junction between two cells, in which information is relayed by direct secretion. Here, for the neuro-immune synapse we are describing a hybrid structure: a specific zone between immune cells and neurons (Shepherd and others 2005) in which both of the systems share common mediators (Shaw and Allen 2001).
When the physiological pain pathway is altered, pain becomes the “enemy” to defeat. In this fight, the role of opioid receptors is still ambiguous, particularly their localization on surfaces of immune cells. A better understanding about peripheral opioid receptors function could be critical for informing more effective CP pharmacological treatment approaches (Machelska and Celik 2020). Clinical studies suggest that opioid administration is associated with high risk of immune suppression and the development of inflammatory mechanisms, enhancing CP status itself (Kosciuczuk and others 2020). Moreover, it is well known that opioid consumption has a dual effect in CP patients, by inducing at the same time tolerance, dependence, and a risk for opioid use disorders (Cameron-Burr and others 2021; Thong and others 2018). Thus, understanding the role of peripheral opioid receptors in CP conditions could also be helpful in minimizing central nervous system opioid side effects, such as addiction, sedation, respiratory depression, and nausea. In this context, we have to approach the problem considering that every single immune cell type could have a role in this process.
Since 1979, several immunomodulatory effects of opioids on T lymphocytes have been described (Sacerdote and others 2003). T lymphocytes, cells of the adaptive immunity, infiltrate the injured central nervous system only after neutrophils and macrophages have already arrived: indeed, macrophages infiltration is essential for lymphocytes recruitment as this event is prevented by macrophages ablation in mice (Ghasemlou and others 2015; Kobayashi and others 2015). T cells interact with neurons in a bidirectional manner. For example, vasoactive intestinal peptide (VIP) induces pro-inflammatory cytokines secretion from CD4+ T-lymphocytes in allergic inflammation. Among these cytokines, secreted interleukin (IL)-5 from lymphocytes in turn enhances neuronal secretion of VIP (Talbot and others 2015). T lymphocytes can stimulate the production of brain-derived neurotropic factor (BDNF) through the action of IL-4 on neurons (Ziv and others 2006). T cells can exert different effects depending on their polarization: Th1 cells, that secrete proinflammatory chemokines (Il-1b, TNF-α, IL.17) enhance pain hypersensitivity while Th2 lymphocytes reduced mechanical allodynia and thermal hyperalgesia in neuropathic models (Moalem and others 2004; Palmer and Weaver 2010). The role of T cells in the induction of pain in inflammatory pain models is poorly understood. For example, T cell–deficient mice do not show a reduction of pain hypersensitivity (Ghasemlou and others 2015; Petrovic and others 2019; Sorge and others 2015) and present a prolonged duration of mechanical allodynia (Laumet and others 2018). In addition, regulatory T cells (T-Reg cells), characterized by the expression of the transcription factor FoxP3 and specialized in dampening inflammation and resolving immune response, can reduce neuropathic pain (Liu and others 2014). Of our interest, T lymphocytes express all the three kinds of opioid receptors on their surface (Liang and others 2016). Human studies have demonstrated that short-term morphine administration induces T-lymphocyte cytokine expression, such as IL-2 and IL-6, by enhancing the differentiation of B-lymphocytes (Campana and others 2010). Work by Campana and others showed that 1-year intrathecal morphine treatment, in chronic non-malignant pain conditions patients, leads to an increase of MOR mRNA level in circulating T-lymphocytes, and this effect is stronger after administration of morphine plus bupivacaine than a pure morphine solution. Assuming a correlation exists between transcription and translation, considerably higher amounts of MOR receptors should be expected in the lymphocytes of patients receiving chronic intrathecal morphine (Campana and others 2010). Moreover, morphine abuse inhibits T-helper 17 (Th17) function and, at the same time, enhances the activity of T-Reg cells. This mechanism could be linked to immune suppression, as reported by Abo-Elnazar and others (2014). Morphine can also increase KOR mRNA expression on T cells, indicating that opioid drugs can exert their effects through multiple opioid receptor subtypes (Suzuki and others 2001). Moreover, activation of DOR receptors expressed on surfaces of T-lymphocytes by the endogenous ligand met-enkephalin enhances the expansion of CD4+ cells and CD4 molecule expression (Shan and others 2011). Today, it is well known that all the elements of the immune system play a specific role in pain. For example, B-lymphocytes, which are antibody-producing cells that represent the source of the humoral immune response, also express μ, δ, and κ opioid receptors (Malafoglia and others 2017). MOR agonists increase IgM and IgG production; DOR agonists produce the opposite effect (Liang and others 2016). Interestingly, B cells are influenced by opioids but the ability to produce antibodies also requires the cooperation of other immune cells, for example, monocytes and macrophages. Monocytes are peripheral blood circulating leukocytes, that rapidly infiltrate the site of infection, an injury, or a damaged tissue. These cells can differentiate toward a pro- or an anti-inflammatory phenotype, depending on the extracellular milieu (Yang and others 2014). Monocytes differentiate into macrophages, typically recognized in Immunohistochemistry sections by the staining with CD68. Chemokines and cytokines contribute to the recruitment of monocytes/macrophages into the peripheral nervous system (Ren and Dubner 2010). In particular, the monocyte chemoattractant protein-1 (MCP-1), also known as CCL2, induces peripheral sensitization acting on CCR2 expressing nociceptors (Zhang and others 2013). Lack of CCR2 in murine models results in reduced macrophages infiltration in nerve injury sites (Siebert and others 2000). Fractalkine, also known as CX3CL1, have been shown to increase in dorsal root ganglia after injury, and blocking CX3CL1 reduces allodynia in paclitaxel-induced neuropathy (Huang and others 2014). TNF-α), a prototypic proinflammatory cytokines, seems to be also involved in macrophages recruitment as it is impaired in TNF-α-deficient mice (Shubayev and Myers 2000). Accordingly, IL-1β accumulates in nerve injury sites and support macrophages recruitment (Perrin and others 2005). The role of macrophages in the induction of pain is demonstrated also in murine models, where macrophages are depleted by the injection of clodronate liposomes: in these mice, thermal and mechanical hyperalgesia are reduced (Liu and others 2000). However, macrophages can sustain tissue repair and regenerative process in several tissue, including nervous tissue (Liu and others 2019), thanks to their capacity to differentiate into M2-anti-inflammatory macrophages. These are immunosuppressive cells, that secrete anti-inflammatory cytokines and growth factors to promote tissue repair and resolution of pain (Mokarram and others 2012). Polarization of macrophages into IL-10 producing M2 cells enhances the resolution of inflammation, thus dampening hyperalgesia (Willemen and others 2014), and in vitro generated M2 macrophage have been shown to secrete opioid peptides, including metenkephalin, dynorphin A, and β-endorphin (Pannell and others 2016). Macrophage cells express opioid receptors themselves. MOR activation by morphine has been shown to regulate macrophages functions, including nitric oxide production and phagocytosis (Brack and others 2004). Macrophage MOR is up-regulated by cytokines, namely IL-1β, IL-4, IL-6, TNF-α, and interferon-γ (IFN-γ). Moreover, an in vitro murine study showed that IFN-γ also stimulates the expression of macrophage’s KOR (Gabrilovac and others 2012). Neutrophils and mast cells are two other major types of immune cells involved in pain modulation. Neutrophils are innate immune cells that rapidly gather around the damaged or injured tissue (Kanashiro and others 2020). Recruitment of neutrophils, induced by T leukotriene B4 (LTB4) and complement component 5a (C5a) is associated with pain sensitization (Ting and others 2008). Neutrophils can act on neurons by inducing the secretion of chemotactic factors that in turn recruit more neutrophils, thus amplifying the nociceptive response (Grace and others 2014). In humans, neutrophils accumulation in the joints of arthritis patients is associated with the induction of hyperalgesia. On the other hand, neutrophils can also secrete analgesic mediators such as opioid peptides (β-endorphin, met-enkephalin, and dynorphin-A), that in turn can inhibit nociceptive transmission by activating opioid receptors on peripheral sensory neurons (Rittner and others 2009). Neutrophils express opioid receptors on their surface and a murine study demonstrated that morphine can completely attenuate neutrophil migration to the site of inflammation and opioids consumption can impede the bactericidal action of these cells (Kanashiro and others 2020; Roy and others 2011). Mast cells are leukocytes with a cytoplasm rich in granules that are resident in tissue, in particular in connective ones. They participate to the immune response to injury by secreting their granules’ content, releasing cytokines, chemokines and other mediators such as histamine (Forsythe and Bienenstock 2012). Mast cell have been shown to increase in human inflammatory diseases (Nigrovic and Lee 2005), and in murine models, which lack mast cells, a reduction of pain can be observed (Milenkovic and others 2007). Blocking histamine signaling by using histamine receptor antagonists reduces mechanical and thermal hyperalgesia (Gupta and Harvima 2018; Liu and others 2021). Mast cells express opioid receptors in their surface but the mechanism that describe the communication between these cells and peripheral nerves in pain pathways is still unknown. A common opioid side effect is the activation of mast cells (Nguyen and others 2014). Interestingly, a human study demonstrated that morphine and other opioids with lower MOR affinity induce mast cells activation; differently MOR potent agonists (i.e., naloxone, buprenorphine) did not activate mast cells. These results indicate that MOR is not directly involved in mast cells activation (Blunk and others 2004). In 2017 a silico design study (Lansu and others 2017) proposed a unique atypical opioid-like receptor impor-tant for modulating mast cell degranulation, named MRGPRX2. Later, a preclinical primate study demonstrated that MRGPRX2 is necessary for innate immune cells recruitment at the injury site, mediating neurogenic inflammation and pain (Green and others 2019). In the past few decades, our group has been focused on MOR expression on natural killer (NK) lymphocytes and lymphokine-activated killer (LAK) cells. These are specific cytotoxic cells of the innate immune system. We demonstrated that, in cancer patients, after chronic in vivo analgesic therapy with morphine, the endogenous cytotoxic activity of NK cells was reduced, while LAK cell cytotoxicity increased. Then, we showed that LAK cell activity mainly increased after an oral morphine administration, rather than an intrathecal one (Provinciali and others 1991). Later, we showed that the effect of morphine on LAK cells activation, but not on NK cell reduction, is related to the modulation of prolactin levels determined by the opioid drug (Provinciali and others 1996). Moreover, we observed an increase of MOR mRNA levels in lymphocytes and a reduction of the percentage of NK cells also in non-cancer-pain patients (CNCP), treated for long time with intrathecal morphine (Campana and others 2010). Interestingly, a recent systematic review (Diasso and others 2020) highlighted the effects of long-term opioids treatment in CNCP, demonstrating that the majority of the articles analyzed had considerable limitations (e.g., cross-sectional design, lack of randomization and/or clinical description, small sample size). According to the authors, very few studies increased our understanding in this field (Campana and others 2010; Tabellini and others 2014), despite that their findings could be not comparable because of diverse opioid formulations and administrations. In addition, although the level of evidence is weak, long-term opioid treatment alters the immune system in CNCP referring not only to NK cells alteration but also to IL-1ß production as a consequence of toll-like receptors (TLRs) stimulation (Dutta and others 2012; Meng and others 2013). In this contest, it has been demonstrated that opioids are TLR4 agonists (Zhang and others 2020b) and an increased IL-1β production has been observed after PBMC stimulation with TLR2 and TLR4 agonists (Kwok and others 2012). These data could be explained and enhanced considering a recent finding (Chang and others 2021), which describes the cross-talk of TLRs and MOR in a preclinical rodent model of chronic constriction injury (CCI). Here the authors suggested that mechanical hyperalgesia might be the result of the cross-talk between TLRs and MOR in a PKCα-dependent manner, opening the way for novel neuropathic pain therapeutic strategies (Table 2). Interestingly, a recent clinical study (Lassen and others 2021) reported findings about the role of NK cells in pain disorders associated with central pain sensitization, (i.e., herpes zoster neuralgia, polyneuropathy). The authors showed that a low NK-cell frequency in the cerebrospinal fluid (CSF) was associated with central sensitization, unlike a high NK-cell frequency, which seemed to prevent it. Thus, a future project could be focused on analyzing opioid receptors expression and pharmacological modulation on NK cells in the CSF. The comparison between central and peripheral analysis could be pivotal to set up the right opioid and/or immunologic treatment.
Recently, inspired by the increasing attention on the role of peripheral opioid receptors in pain pathways, our group decided to analyze the percentage of immune cells expressing opioid receptors in CP patients suffering from fibromyalgia or osteoarthritis. Interestingly, we found that the percentage of B cells expressing MOR was lower in CP patients than in a pain-free control group. This difference was greater in CP patients with severe pain. Thus, for the first time, MOR could be considered as a potential peripheral CP biomarker (Mu-Lympho-Marker, MLM) (Fig. 5), considering that B cells express it in patients with FM and OA (Raffaeli and others 2020). The role of MOR as a marker of pain has been also postulated in an ongoing clinical trial (Malafoglia and others 2017), but the meaning of its immunological characteristic is still a topic of discussion. The low percentage of B cells expressing MOR in CP patients could be due to a reduction of an opioid receptor “reserve,” necessary for the pain inhibition pathway mediated by immune cells. In this study, we did not enroll patients taking opioids. Thus, we can exclude a reduction of the opioid receptor reserve as a consequence of desensitization and/or internalization of MOR (Zhang and others 2020a). Still, several questions are open. There could be additional mechanisms of action, and a new study to first assess CP patients could be greatly helpful to set up the right therapeutic strategy. In particular, once we understand the mechanistic role of the MOR reserve reduction, it could be possible to develop peripheral pharmacological treatments based on the ideal opioid dosage administration for each single CP patient, bypassing central side effects. Moreover, future analysis should address how gender and sex differences affect the opioids and immune system interaction in patients suffering from CP. In fact, preclinical and clinical studied reported that women, in general, present a higher immune response than men (Schwarz and Bilbo 2012). Importantly, morphine has a stronger analgesic effect in males than in females (Doyle and Murphy 2017). Sex differences in CP are probably underestimated, although it is well known that women are more likely to use anti-inflammatory drugs, which enhance opiate action (Li and others 2021), than men. Thus, even if more data is needed to confirm the MOR reserve hypothesis, preliminary findings could be helpful to also discriminate gender-dependent CP biomarkers, underlining the importance of peripheral opioid receptors in analgesia and paving the way for new peripheral, tailored pharmacological approaches and rehabilitation strategies for CP patients. In conclusion, the review of the current literature seems to suggest that the identification of specific pain biomarkers remains perhaps the most important challenge in the field of CP medicine (Fig. 3). |
PMC9647332 | Kei Takada,Ryoko Nakatani,Emiko Moribe,Shizuka Yamazaki-Fujigaki,Mai Fujii,Masayo Furuta,Hirofumi Suemori,Eihachiro Kawase | Efficient derivation and banking of clinical-grade human embryonic stem cell lines in accordance with Japanese regulations | 06-11-2022 | Human embryonic stem cells,Clinical grade,Cell banking,hESCs, human embryonic stem cells,GMP, good manufacturing practice,QC, quality control,LM, laminin,CPF, cell processing facility,MHLW, Ministry of Health, Labour, and Welfare,ICM, inner cell mass | Introduction We recently established clinical-grade human embryonic stem cell (hESC) line KthES11 in accordance with current good manufacturing practice standards in Japan. Despite this success, the establishment efficiency was very low at 7.1% in the first period. Methods To establish clinical-grade hESC lines, we used xeno-free chemically defined medium StemFit AK03N with the LM-E8 fragments instead of feeder cells. The protocol was then optimized, especially in the early culture phase. Results We established five hESC lines (KthES12, KthES13, KthES14, KthES15, and KthES16) with 45.5% efficiency. All five hESC lines showed typical hESC-like morphology, a normal karyotype, pluripotent state, and differentiation potential for all three germ layers. Furthermore, we developed efficient procedures to prepare master cell stocks for clinical-grade hESC lines and an efficient strategy for quality control testing. Conclusions Our master cell stocks of hESC lines may contribute to therapeutic applications using human pluripotent stem cells in Japan and other countries. | Efficient derivation and banking of clinical-grade human embryonic stem cell lines in accordance with Japanese regulations
We recently established clinical-grade human embryonic stem cell (hESC) line KthES11 in accordance with current good manufacturing practice standards in Japan. Despite this success, the establishment efficiency was very low at 7.1% in the first period.
To establish clinical-grade hESC lines, we used xeno-free chemically defined medium StemFit AK03N with the LM-E8 fragments instead of feeder cells. The protocol was then optimized, especially in the early culture phase.
We established five hESC lines (KthES12, KthES13, KthES14, KthES15, and KthES16) with 45.5% efficiency. All five hESC lines showed typical hESC-like morphology, a normal karyotype, pluripotent state, and differentiation potential for all three germ layers. Furthermore, we developed efficient procedures to prepare master cell stocks for clinical-grade hESC lines and an efficient strategy for quality control testing.
Our master cell stocks of hESC lines may contribute to therapeutic applications using human pluripotent stem cells in Japan and other countries.
Since the first hESCs were established in 1998 [1], their unique characteristics and potential for the development of regenerative medicine have fascinated many people. In Japan, we had established five hESC lines by 2008 [2,3] in a specifically designed facility, but not in keeping with current good manufacturing practice (GMP). The cell lines were established and maintained using feeder cells such as mouse embryonic fibroblasts and/or SL10 cells, a STO subline. Additionally, we used culture medium containing animal components. Thus, we aimed to establish new hESC lines under GMP-grade culture and operation systems for clinical use. Therefore, we built a cell processing facility (CPF) and prepared to establish clinical-grade hESCs, including the preparation of standard operating procedures. Approximately 10 years after establishing hESC lines in 1998, Crook et al. in Singapore reported six clinical-grade hESC lines derived from embryos imported from Australia [4]. However, they still used animal products for the derivation. Since then, clinical-grade hESCs have been established in many countries, including the United States [5], United Kingdom [[6], [7], [8]], Israel [9], Finland [10], and China [11,12]. Clinical studies using human pluripotent stem cells (hPSCs) have been conducted in at least 10 countries, and significantly in the United States and China. By the end of 2019, there were at least 54 clinical studies of using hPSCs for the treatment of 22 diseases. Of these studies, 32 clinical studies are performed with hESC-derived cell products [13]. In Japan, “The Act on the Safety of Regenerative Medicine” and “Pharmaceuticals, Medical Devices, and Other Therapeutic Products Act” were enacted in November 2014 to realize and commercialize regenerative medicine [14]. Following these laws, it was possible to establish hESC lines for clinical use in Japan. To establish hESC lines for clinical use, we chose E8 fragments of laminin (LM-E8s) instead of feeder cells. Rodin et al. and we showed that laminin-511/521 are a suitable alternative culture substrate to support hPSC culture compared with other extracellular matrix proteins and Matrigel [15,16]. Furthermore, we have demonstrated that LM-E8s provide superior adhesion for hPSC culture compared with intact laminin-511 and efficiently expand hPSCs with high quality and homogeneity [17]. We started to establish clinical-grade hESCs in 2017 and reported the first clinical-grade hESC line, KthES11, in 2018 [18]. During this period, the establishment efficiency was low at one-fourteenth. We had decided on items for quality control (QC) testing of hESCs for clinical use and were performing the tests, but when it was appropriate to perform QC testing was unclear. In this study, we developed a system to establish clinical-grade hESC lines with high efficiency. Furthermore, we performed QC tests separately during the manufacturing process for master cell stocks.
The study was approved by the ethics committee of Life and Medical Sciences, Kyoto University (approval number, ES-1), for derivation and characterization, including early differentiation of hESC lines from donated blastocysts, where informed consent was obtained from donors’ parents. Additionally, the study was approved by the Ministry of Education, Culture, Sports, Science, and Technology (license number: 0630) and the Ministry of Health, Labour, and Welfare (MHLW) (license number: 0630) in Japan.
Derivation of clinical-grade hESC lines and their banking were performed in our CPF operated with GMP-level management. In line with “The Act on the Safety of Regenerative Medicine,” the facility obtained a manufacturing license (#FA5160004) from MLHW after passing a site visit by the Pharmaceuticals Medical Devices Agency in Japan. Long-term culture and QC tests of our master cell stocks were performed outside the CPF.
We gently isolated the inner cell mass (ICM) from day 5–6 blastocysts using a micromanipulator (Narishige, Tokyo, Japan). The ICMs were seeded on a Falcon Center Well Organ Culture Dish (Corning, NY, USA) precoated with LM-E8s (0.5 μg/cm2, iMatrix-511MG, Matrixome, Osaka, Japan) in StemFit AK03N (Ajinomoto, Tokyo, Japan) with Y-27632 (final concentration: 10 μM; FUJIFILM Wako, Osaka, Japan). Although ICMs had fully attached by the next day, we used Y-27632 during ICM culture for the initial several passages. At approximately 7 days, ICMs with outgrowths were mechanically dissected into small pieces and transferred into a new culture well precoated with iMatrix-511MG. After 14–16 days of culture in an incubator at 37 °C with 5% CO2 and 5% O2, morphologically hESC-like cells had appeared. The cells were subsequently cultured for three to four passages by mechanical dissection using 31 G needles, and the resulting cells appeared to be stabilized as hESCs. At this stage, the hESCs were capable of passaging as small clumps using a non-enzymatic solution (5 mM EDTA in PBS) and cryopreservation for early stocks. When seeded at 1–2 × 105 cells/well (6-well plate), the cells became confluent every 3–4 days. During culture, cell viability was examined at each passage by a Via1-Cassette™ with NucleoCounter NC-200 (ChemoMetec, Allerød, Denmark). Cell viability was generally higher than 75%.
hESC lines at early passages were harvested in a batch of five to 10 vials as seed stocks with 5 × 105 to 2 × 106 viable cells per vial. To prepare master cell stocks, we thawed one vial from seed stocks, expanded the cells in culture, and then banked the cells in batches of 50 vials with >1 × 106 viable cells per vial in cryopreservation medium consisting of StemFit AK03N with 10% DMSO (Sigma–Aldrich, St. Louis, MO, USA) and Y-27632 (final concentration; 5 μM). The number of viable collected cells was determined by double staining with acridine orange and DAPI (ChemoMetec). The vials were labeled for traceability and placed in a CoolCell freezing container (Biocision, San Rafael, CA, USA) in a −80 °C mechanical freezer (PHC, Tokyo, Japan). The next day, the vials were transferred to the vapor phase of a liquid nitrogen tank (LN2 supply tank, Taiyo Nippon Sanso, Tokyo, Japan).
Current drafts for QC testing are described in Table 1. Our QC testing was based on the guidelines of the International Stem Cell Banking Initiative [19]. We used four categories: (1) in-process, (2) cell authenticity, (3) biological safety, and (4) characterization and potency. For example, we performed photomicrography daily to confirm that >80% of colonies were undifferentiated. When hESCs were subcultured, we examined the total cell number and viability. In general, cell viability was >75%. As cell authenticity tests, HLA and STR tests were performed during master cell stock preparation. For biological safety tests, sterility tests were performed during both seed stock and master cell stock preparations. Other biological safety tests were performed during master cell stock preparation. Conversely, characterization and potency were examined as long-term culture tests from more than three independent trials using our master cell stocks. Karyotype analysis was performed using seed and master cell stocks, and after every five passages during long-term culture tests. Detailed QC testing protocols have been described previously [8]. For ectodermal differentiation, we used STEMdiff™ Trilineage Ectoderm Medium (STEMCELL Technologies, Vancouver, Canada) in accordance with the manufacturer's instructions except that the number of induction days was extended from 7 to 11 days. For neuronal cell differentiation, we used a Quick-Neuron™ Dopaminergic - mRNA Kit (Elixirgen Scientific, Baltimore, MD USA) in accordance with the manufacturer's instructions. Then, the cells were fixed in 4% paraformaldehyde, permeabilized with 0.2% Triton X-100 in PBS, and stained with the primary and secondary antibodies described in Supplementary Table 1. We performed antibody incubations for 1 h at room temperature or overnight at 4 °C. Nuclei were counterstained with DAPI (Thermo Fisher Scientific). ABO genotyping was performed by allele-specific PCR as described previously [20] with the following modifications. Genomic DNA from each cell line was purified using a FlexiGene DNA Kit (Qiagen) in accordance with the manufacturer's instructions. Real-time PCR using 1 ng DNA was carried out with Power SYBR Green Master Mix (Thermo Fisher Scientific) and the StepOnePlus™ Real-Time PCR System (Thermo Fisher Scientific). Thermal cycling was denaturation at 95 °C for 10 min, followed by 40 cycles of 95 °C for 10 s, 60 °C for 20 s, and 72 °C for 30 s.
To prepare stocks efficiently in a compact space with a small number of people, we developed a system for clinical-grade hESC banking as shown in Fig. 1, which was characterized by the following points. 1. Six to 10 weeks were required to produce master cell stocks from 5–6-day blastocysts. Therefore, we prepared seed stocks before the master stock and divided the manufacturing process into two parts to allow for flexibility in the manufacturing schedule. We generally prepared five to 10 frozen vials as seed stocks. 2. We performed QC tests separately in stages as shown in Table 1 and Fig. 1. Cell authenticity and biological safety were examined during master cell stock preparation in the CPF. Conversely, other characterizations, such as hESC marker expression and differentiation potential, were performed using long-term cultured cells from master cell stocks with thawing (QC test 3). 3. Long-term culture tests were performed in a specially designated facility with an air conditioning system with a HEPA filter. All materials and solutions were the same grades as those in the CPF. 4. Before master cell stock preparation, we continuously cultured some hESCs from seed stock preparation outside the CPF and developed optimal culture protocols for each cell line. Furthermore, we performed some QC tests, such as gene expression analysis and flow cytometry, to confirm that seed stocks had some hESC characteristics in addition to their morphology. 5. Karyotype analyses were carried out during master cell stock preparation and every five passages during long-term culture.
We established only one clinical-grade hESC line, KthES11, from 14 blastocysts in period 1. One major problem was poor adhesion of the ICM to LM-E8s. We previously showed that LM-E8s promote strong adhesion of hPSCs compared with other culture substrates, such as Matrigel and intact laminin [17], suggesting that the remaining trophectoderm cells were not suitable for adhesion to LM-E8s. Therefore, we carefully removed the trophectoderm area from blastocysts using micromanipulators. As expected, proper ICMs adhered well to LM-E8s until the next day. Another major issue was the sensitivity against subculturing hESC-like cells during initial passages, even when we used non-enzymatic 5 mM EDTA/PBS and added Y-27632 to the culture medium. Therefore, we applied mechanical dissociation using 31 G needles for the stages. After hESC-like cells had appeared to stabilize with a sufficient number (passage 3 to 4), the cells were passaged as small colonies using 5 mM EDTA in PBS as described in the Materials and Methods. With these improvements, our establishment efficiency was increased to 45.5% (five out of 11) in Period 2 from 7.1% (one out of 14) in Period 1 (Table 2). Furthermore, we established clinical-grade hESCs using iMatrix-511MG at a concentration of 0.5 μg/cm2, but the cells could be generally maintained using iMatrix-511MG at a lower concentration (e.g., 0.2–0.25 μg/cm2). At this concentration, cell viability during subculture was approximately 90%.
In the second period, we established another five clinical-grade hESC lines of both sexes with different blood types (Table 2). We had already reported the KthES11 line [18], but had not evaluated the cells in long-term culture. Therefore, we evaluated the cells in conjunction with the five hESC lines established in this study. All hESC lines showed typical hESC-like morphology (Fig. 2). Flow cytometric analysis also revealed high expression of pluripotent cell markers of transcriptional factors (NANOG, OCT4, and SOX2) and cell surface markers (TRA-1-60, TRA-1-81, and SSEA-4) (Fig. 3). Cell surface marker SSEA-3 showed moderate to high expression depending on the cell line. Conversely, the differentiation marker SSEA-1 showed low expression. For karyotype analysis, we performed G-banding every five passages from our seed stocks. All six cell lines showed normal karyotypes in long-term culture (Fig. 3). To assess pluripotency, hESCs in long-term culture were grown in suspension in custom mTeSR™1 (without bFGF and TGFb; STEMCELL Technologies). The cells formed embryoid bodies (EBs) (Fig. 4A), and then the EBs expressed markers of all three germ layers as confirmed by the TaqMan hPSC Scorecard Panel (Thermo Fisher Scientific) (Fig. 4B). The score box plot shows that the expression of markers for all three germ layers derived from the six clinical-grade hESC lines was increased significantly, except for KthES13 cells that had elevated expression of ectoderm markers, but not at a significant level (Fig. 4C). Therefore, we examined ectoderm differentiation in KthES13 cells using STEMdiff™ Trilineage Ectodermal Medium (STEMCELL Technologies). In this system, the score of ectoderm markers became significant (Supplementary Figs. 1A and B). Furthermore, KthES13 cells differentiated into dopaminergic neurons using a differentiation kit (Elixirgen Scientific) (Supplementary Fig. 1C). Taken together, we concluded that KthES cells had a differentiation potential for all three germ layers. Our master cell stocks of all clinical-grade hESCs were sterile, free of mycoplasma, and negative for the viruses described in the Materials and Methods. Furthermore, the cell lines passed an endotoxin test. These data are available from the corresponding author.
We established five clinical-grade hESC lines without feeders at high efficiency. Furthermore, we developed an efficient system for QC testing. All five hESC lines were pluripotent with normal karyotypes. The hESC lines also showed a differentiation potential in an EB formation assay and other assessment methods. ICM is inside the blastocyst and covered with trophectoderm, which expresses integrin αVβ3 [21,22]. Accordingly, fibronectin or vitronectin is suitable to adhere the trophectoderm, but not laminin. Therefore, we isolated ICMs by gentle removal of the trophectoderm, and then most ICMs attached well. We used a non-enzymatic EDTA solution for subculture. This method does not dissociate cells into single cells and appears to be less damaging than dissociation by enzymatic solutions such as TrypLE Select (Thermo Fisher Scientific). Although there was good cell viability with normal passaging, cell viability was often low in the first phase of the establishment process. The preparation of master cell stock was often very slow. Therefore, we dissociated the cells mechanically when cultured in a center well organ culture dish, i.e., when the number of cells was small. These changes greatly improved the efficiency of establishment. Our aims when establishing clinical-grade hESC lines consisted of three evaluation stages: preparation of seed stocks, preparation of master cell stocks, and evaluation by long-term culture using the master cell stocks. We did not perform all QC tests for every step, but they were assigned to each stage. When we prepared seed stocks, we performed sterility tests for biological safety. We also cultured the cells outside of the CPF using surplus cells from cryopreservation. We confirmed that the cell lines had a normal karyotype by G-binding. We examined whether the cell stocks had hESC characteristics by FACS analysis and immunocytochemistry. The reason why the established cells were hESCs was because of their morphology. Additionally, we optimized the culture method. For some cell lines, the concentration of iMatrix-511MG was lowered to 0.25 μg/cm2 to facilitate cell detachment and increase cell viability in passages. These pretests led to efficient preparation of the master cell stocks and QC testing. Next, when preparing the master cell stocks, biological safety tests were the focus. In period 1, we used antibiotics in the medium during early culture. Therefore, we examined antibiotic residues using the highly sensitive LC/MS/MS method. Subsequently, we did not add antibiotics and did not need to conduct such a test. Evaluation tests for long-term culture were conducted outside of the CPF in a specially designated facility with an air conditioning system that used HEPA filters, while the reagents used for culture were the same as those used in the CPF. We have frozen and stored some of the master cell stocks and long-term cultured cells as evaluation samples for further development of QC testing. Currently, we are distributing these clinical-grade hESC lines upon request. We expect to reveal more characteristics of our hESC lines through collaboration with our distribution partners. For example, we may find unique characteristics of the cell lines that can easily differentiate into specific cell types, cell lines that maintain genome stability even after long-term culture, or cell lines suitable for genetic recombination. Because of the great potential of hESCs, many hESC lines have been established worldwide and have been registered in hPSCreg (https://hpscreg.eu/search?cell-type=hesc), in which we reported our cell lines, including those in this study. However, in Japan, only two centers, our institute and the National Research Institute for Child Health and Development [23], have succeeded in establishing hESC lines. Therefore, we have been entrusted with an important task: to establish hESC lines in Japan.
We established hESCs for clinical use with high efficiency. Our master cell stocks of the hESC lines may contribute to therapeutic applications using hESCs in Japan and other countries.
The authors declare no competing interests. |
|
PMC9647352 | Anna He,Cheng Fang,Yue Ming,He Tan,Mengyi Zhang,Ruiqing Zhang,Jingyi Li,Mingzhu Nie,Fengyu Li,Yaxin Hu,Xinxin Shen,Xiuge Rong,Xuejun Ma | Development of field-applicable endogenous internally controlled recombinase-aided amplification (EIC-RAA) assays for the detection of human papillomavirus genotypes 6 and 11 using sample releasing agent | 02-11-2022 | Human papillomavirus 6 and 11,Human β-globin gene,Endogenous internally controlled recombinase-assisted amplification,Sample releasing agent | Objective Human papillomavirus (HPV) 6 and 11 are the two most common low-risk HPV subtypes, accounting for more than 90% of condyloma acuminatum. A simple, accurate and rapid screening method to be applied in community-level hospitals is in high demand. Methods Endogenous internally controlled recombinase-assisted amplification (EIC-RAA) assays for HPV6 and 11 were performed in a single closed-tube at 39 °C within 30 min. The sensitivity and specificity of EIC-RAA were examined using recombinant plasmids and pre-tested HPV DNA. A total of 233 clinical samples were collected, and the DNA was extracted by traditional multi-step extraction, or sample releasing agent, before analysis by EIC-RAA. For comparison, HPV detection via Quantitative real-time PCR (qPCR) was also performed. Results The sensitivity of EIC-RAA analysis was 10 copies/reaction for HPV6, 100 copies/reaction for HPV11, and 100 copies/reaction for the human β-globin gene. No cross-reaction was observed with other HPV subtypes. Clinical performance of the EIC-RAA assay achieved a 100% of concordance rate with the commercial HPV qPCR kit. Further, the EIC-RAA assay achieved a 100% of concordance rate when using multi-step extracted DNA and sample releasing agent-processed DNA. Summary The EIC-RAA assay for HPV6 and 11 detection possesses the advantages of accuracy, simplicity and rapidity, and demonstrates great potential to be used in community-level hospitals for field investigation. | Development of field-applicable endogenous internally controlled recombinase-aided amplification (EIC-RAA) assays for the detection of human papillomavirus genotypes 6 and 11 using sample releasing agent
Human papillomavirus (HPV) 6 and 11 are the two most common low-risk HPV subtypes, accounting for more than 90% of condyloma acuminatum. A simple, accurate and rapid screening method to be applied in community-level hospitals is in high demand.
Endogenous internally controlled recombinase-assisted amplification (EIC-RAA) assays for HPV6 and 11 were performed in a single closed-tube at 39 °C within 30 min. The sensitivity and specificity of EIC-RAA were examined using recombinant plasmids and pre-tested HPV DNA. A total of 233 clinical samples were collected, and the DNA was extracted by traditional multi-step extraction, or sample releasing agent, before analysis by EIC-RAA. For comparison, HPV detection via Quantitative real-time PCR (qPCR) was also performed.
The sensitivity of EIC-RAA analysis was 10 copies/reaction for HPV6, 100 copies/reaction for HPV11, and 100 copies/reaction for the human β-globin gene. No cross-reaction was observed with other HPV subtypes. Clinical performance of the EIC-RAA assay achieved a 100% of concordance rate with the commercial HPV qPCR kit. Further, the EIC-RAA assay achieved a 100% of concordance rate when using multi-step extracted DNA and sample releasing agent-processed DNA.
The EIC-RAA assay for HPV6 and 11 detection possesses the advantages of accuracy, simplicity and rapidity, and demonstrates great potential to be used in community-level hospitals for field investigation.
The human papillomavirus (HPV) includes more than 200 different subtypes, divided into the high- and low-risk categories according to their biological characteristics and carcinogenic potential (Zhu et al., 2019). Although the high-risk subtypes are the main pathogenic agents, the low-risk subtypes can cause condyloma acuminatum and low-grade cervical intraepithelial neoplasia, which still pose a considerable health burden to people's lives. HPV6 and HPV11 are the two most common low-risk subtypes, accounting for more than 90% of condyloma acuminatum (Sturegard et al., 2013). Condyloma acuminatum has become a secondary sexually transmitted disease in sexually active adults, with about 1 million new cases every year in the world (Yu et al., 2018). Moreover, about one-third of condyloma acuminatum cases experience relapse after treatment, resulting in huge medical costs, serious psychological and physical burden to patients. Epidemiological data show that subclinical infection, with no obvious symptoms, is the main form of HPV6 and 11 infections, and is closely related to the high recurrence rate of condyloma acuminatum (Flores-Diaz et al., 2017). Therefore, early, simple and rapid screenings of HPV6 and HPV11 infection and timely treatment of condyloma acuminatum are in high demand, particularly in community-level hospitals. At present, culturing HPV in vitro remains challenging. Owing to the lack of in vitro culture systems for etiological detection and reliable immunological assays, the detection of HPV is almost entirely dependent on molecular assays (Jamshidi et al., 2012). Currently, the way to test for HPV6 and HPV11, including genotype specific polymerase chain reaction (PCR), Quantitative real-time PCR (qPCR), polymerase chain reaction-restricted fragment length polymorphisms (PCR-RELP), droplet digital PCR (ddPCR), loop-mediated isothermal amplification (LAMP) and hybrid capture 2 test (HC2) (Table 1). these kits possess high specificity and sensitivity, but require high skilled technical personnel, high cost, long detection time, well-equipped laboratories, cumbersome temperature-control devices and sophisticated analysis, which limit the wide adoption of these methods in resource-poor areas (Grce et al., 2000; Hawkins et al., 2013; Lindh et al., 1992; Oliveira et al., 1994; Yu et al., 2015). Therefore, it is essential to develop a more efficient method for rapid screening of HPV6 and HPV11 infections to promote early detection and treatment. Recombinase-aided amplification (RAA) is an isothermal amplification reaction containing recombinant enzymes extracted from bacteria or fungi, single-stranded DNA binding (SSB) protein, DNA polymerase and specific primers. At an isothermal temperature of 39 °C, the recombinant enzymes can tightly combined with the primers to form the polymer of the enzyme and the primer. With the help of SSB protein the primer searches the template DNA for a complete match of the complementary sequence, the double-stranded structure of the template DNA then is opened. Under the action of DNA polymerase, a new DNA complementary chain is formed. The RAA detection process can usually be completed in less 30 min. In our previous studies, RAA has been successfully applied to the detection of various pathogens and bacteria (Bai et al., 2020; Shen et al., 2019; Wang et al., 2019; Zhang et al., 2019). However, the method required addition of an exogenous reference plasmid as an internal control, which also competed with the target gene. In this study, we further optimized the RAA method for HPV detection by utilizing the human β-globin gene as an endogenous control, along with specific HPV6 and HPV11 probes. Addition of sample releasing agent further simplifies the sample extraction method and improves the utility of this method in common clinical settings. The endogenous internally-controlled RAA (EIC-RAA) assay was compared with commercial HPV detection kits to evaluate the feasibility of the EIC-RAA assay for field use.
A total of 191 cervical exudative cell samples, and 42 condyloma acuminatum swab samples were collected from Tangshan Gongren Hospital (Hebei, China) and Hebei General Hospital (Hebei, China). The average age of patients was 32.32 ± 8.55 years old. HPV DNA typing kit (qPCR method, Sansure, Hunan, China) was used to screen these samples for 26 HPV subtypes. Among them, 62 cases were positive for HPV6, 12 cases were positive for HPV11, and 8 cases were positive for both HPV6 and HPV11. The remaining 118 cases were positive for other HPV subtypes (HPV16, 18, 31, 33, 35, 39, 40, 42, 43, 44, 45, 51, 52, 53, 54, 55, 56, 57, 58, 59, 66, 67, 68 and 73), and 33 samples were negative for HPV. This study was approved by the Institutional Review Boards of the National Institute for Viral Disease Control and Prevention, Center for Disease Control and Prevention of China. The patients gave informed consent to the purpose of the study.
DNA was extracted from clinical samples using viral RNA/DNA isolation kits (Tianlong, Suzhou, China) in strict accordance with the manufacturer's instructions. The extracted DNA was stored at −80 °C until further use. In parallel, the samples were treated with sample releasing agent kit (Sansure, Hunan, China) as follows: 200 μL of concentrated solution and 200 μL of the original raw sample were mixed and centrifuged at 12,000 rpm for 10 min. After discarding the supernatant, 50 μL of nucleic acid releasing agent was added and left to stand for 10 min. The prepared nucleic acids were kept at 4 °C, and preferably used on the same day.
Thirty complete genome sequences of HPV6, HPV11 and human β-globin were downloaded from NCBI GenBank (Table 2) Vector NTI (Thermo Fisher Scientific, USA) was used to align their gene sequences separately, and the Oligo7 (Molecular Biology Insights, USA) software was used to design probes and primers for the highly conserved gene regions of HPV6, HPV11, and human β-globin, according to the design principles of RAA (Bai, 2020). The specificity was further assessed by BLAST on the NCBI website. The finalized primers and probes listed in Table 3 were synthesized by Sangon (Shanghai, China).
A 351 bp length (NT6200-6550 bp, GenBank accession No. AF092932.1) of HPV6, a 391 bp length (NT5950-6350 bp, GenBank accession No. M14119.1) of HPV11, and a 300 bp length (NT352-651 bp, GenBank accession No. MK476504.1) of human β-globin were cloned into the pUC57 vector. The plasmid DNA concentration was quantified using a Qubit 2.0 fluorometer and Qubit dsDNA BR Assay Kit (Life Technologies, Warrington, UK). The DNA concentrations were converted to genome copies using the following formula: DNA copy number (copy number/μL) = [6.02 × 1014 DNA concentration (ng/μL) × 10− 9]/[DNA length in nucleotides × 660] (Ma et al., 2017). The plasmids were diluted 10-fold from 105 to 101 copies/μL, and stored at −20 °C.
Singleplex real-time RAA tests for HPV6, HPV11 and human β-globin were performed separately in reaction volumes of 50 μL using the RAA EXO kit (Qitian, Jiangsu, China). The reaction components included 25 μL of buffer, 16.7 μL of DNase free water, 2.1 μL of HPV6, HPV11, or human β-globin gene forward and reverse primers (10 μM) respectively, 0.6 μL of HPV6, HPV11 or human β-globin gene-specific RAA probe (10 μM), and 1 μL of DNA template or 1 μL of DNase free water. A 47.5 μL of the reaction mixture was added to the lyophilized RAA particle kit containing all necessary enzymes (SSB, 800 ng/μL; UvsX, 120 ng/μL; DNA polymerase, 30 ng/μL), and 2.5 μL of 280 mM magnesium acetate was transferred into the tube. The capped tube was placed in the RAA-B6108 instrument for 4 min of pre-amplification at 39 °C, followed by 30 min in the QT-RAA-F1620 real-time fluorescence detection system (Qitian, Jiangsu, China), at 39 °C for detection. Positive results were determined by the equipment QTRAA-F1620 by setting the slope K value to be greater than or equal to 20 and was simultaneously recorded in the FAM and HEX detection channels. A serial dilution of the recombinant plasmid to 105, 104, 103, 102, 101 copies/μL, was used to determine the sensitivity of singleplex RAA for HPV6, HPV11 and human β-globin. The negative control reaction was carried out in parallel with each experiment, The specificity of the RAA test was evaluated with cross-reactivity to 118 other HPV positive samples.
Standard recombinant plasmids of HPV6, HPV11, and human β-globin were prepared in the dilution range of 101–105 copies/μL. In order to optimize the EIC-RAA analysis so that the amplification of human β-globin in the HEX channel has minimal influence on the sensitivity of the amplification of target genes in the FAM channel, the target plasmids of 101–105 copies were tested in the presence of 101–105 copies of human β-globin to examine weather the human β-globin could interfere with detection of HPV6 and HPV11. The optimized EIC-RAA assay was then performed in a 50 μL reaction volume consisting of 25 μL of rehydration buffer, DNase-free water, 2.1 μL of target-specific forward and reverse primers (10 μM), 0.6 μL of HPV6 or HPV11 probe (10 μM), human β-globin probe (10 μM), human β-globin specific forward and reverse primers (10 μM) and 2.5 μL of 280 mM magnesium acetate.
A sample releasing agent kit (Sansure, Hunan, China) consists of concentrated solution and nucleic acid releasing agent. They were used to concentrate the sample, denature proteins, and release nucleic acid. As the use of different concentrated solution and sample input volumes may affect the efficiency and the applicability to the EIC-RAA experimental method, we tested the EIC-RAA assay under different working conditions. A total of 100 μL, 200 μL or 300 μL of concentrated solution was mixed with 100 μL or 200 μL of sample, respectively, into a 1.5 mL EP tube and centrifuged at 12,000 rpm for 10min. After discarding the supernatant, 50 μL of nucleic acid releasing agent was added and left to stand for 10 min. One to five microliters of the reaction mixture was then tested by EIC-RAA analysis to determine the optimal template input volume. The optimal reaction conditions were determined by the time threshold and fluorescence measured by EIC-RAA.
EIC-RAA and qPCR were used to detect nucleic acids in 233 samples. The HPV nucleic acid detection kit (PCR-fluorescence probe method) was purchased from Sansure (Hunan, China) and was able to detect 26 HPV subtypes and identify specific genotypes (HPV6,11,16, 18, 31, 33, 35, 39, 40, 42, 43, 44, 45, 51, 52, 53, 54, 55, 56, 57, 58, 59, 66, 67, 68 and 73). qPCR reactions were performed according to the manufacturer's instructions using a CFX96 qPCR System (BIO-RAD, USA).
IBM SPSS Statistics, version 21 (IBM Corporation, NY, USA) was used to perform all of the statistical analysis. The results were analyzed using kappa tests, and a P-value less than 0.05 was considered statistically significant.
The sensitivity of HPV6, HPV11 and human β-globin was as low as 10 copies/reaction, and no abnormality was found in the negative control. Moreover, the results also showed no cross-amplification with other HPV genotypes (HPV16, 18, 31, 33, 35, 39, 40, 42, 43, 44, 45, 51, 52, 53, 54, 55, 56, 57, 58, 59, 66, 67, 68 and 73; data not shown).
The concentration of internal gene has different effects on the amplification efficiency of target genes and different concentrations of target gene have different effects on the amplification of internal gene beacuse of the co-consumption of RAA reagents in the reactions, The sensitivity of target gene was 10 copies/reaction for HPV6 (Figure 1.1A), 100 copies/reaction for HPV11 (Figure 1.2A) in the presence of 105 copies human β-globin plasmid, indicating the internal gene did not significantly affect the amplification sensitivity of the target gene, while the internal gene was steadily detected in the presence of target gene with a concentration range of 101–105 copies/reaction in the HEX channel (Figure 1.1B and 1.2B).
The amplification curves of the combination of 200 μL of concentrated solution, 200 μL of sample, and 50 μL of nucleic acid releasing agent showed the shortest threshold time (the initial peak time of the fluorescence curve) and the highest fluorescence value to obtain the best amplification efficiency for HPV6 (Figure 2.1A) and HPV11 (Figure 2.1B). Similarly, the best amplification efficiency of EIC-RAA was obtained by loading 2 μL of reaction mixture (Figure 2.2A and B).
qPCR detection gave Ct values ranging from 17.01 to 37.15 for HPV6, 20.50 to 36.48 for HPV11, and 19.68 to 37.86 for human β-globin (according to the instruction of Sansure kit, qPCR-positive Ct cut off value was set to 39). For 233 previously defined samples, the comparison result between EIC-RAA and qPCR detection of HPV6, HPV11 and human β-globin, were shown in Table 4. The concordance rate between EIC-RAA assay and RT-PCR was 100%, with kappa value of 1 (P < 0.01). Among these condyloma acuminatum swab samples, the internal reference, human β-globin, was not detected in 3 specimens. For the EIC-RAA assay, the coincidence rate between target genes and internal reference was 100%, whether samples were treated with sample releasing agent or multi-step DNA extraction, and the Kappa value was 1 (P < 0.01).
In this study, we established the EIC-RAA assay for the accurate detection of HPV6 and HPV11, using the human β-globin gene as internal control. To our knowledge, this is the first study to detect HPV6 and HPV11 using RAA technology. The EIC-RAA test for HPV6 and 11 was proved to be highly specific, as no cross-reactivity with other HPV subtypes was observed. A total of 233 specimens were assessed by EIC-RAA and qPCR. The results of EIC-RAA and qPCR were consistent. When DNA extractions of the samples were performed with sample releasing agent versus viral RNA/DNA Isolation Kit, the results of the two methods were consistent. Sample releasing agent can greatly simplify the steps of nucleic acid extraction, reduce the use of large-scale instruments, shorten the nucleic acids extraction time, and make EIC-RAA detection more suitable for screening in resource-poor areas. The EIC-RAA assay is a simple and rapid isothermal amplification technique, and the amplification results can be obtained at 39 °C in 30 min. The sensitivity of EIC-RAA was 10 copies/reaction for HPV6 and 100 copies/reaction for HPV11. In comparison, HPV-LAMP, was previously reported to detect HPV6 and HPV11 with a sensitivity of 1 000 copies/reaction, and requiring 2 h (Hagiwara et al., 2007; Zhong et al., 2018). Furthermore, HPV-LAMP requires 4 primers, which is more complicated than the RAA primer design. The EIC-RAA assay is therefore superior to HPV-LAMP in terms of time and sensitivity. As for commercial RT-PCR kits, which are widely used in the clinics, the reported sensitivity is in the range of 44–200 copies/reaction (Kocjan et al., 2008; Micalessi et al., 2011; Wang et al., 2021). Thus, the EIC-RAA assay is more sensitive for the detection of HPV6 and HPV11. Human β-globin is expressed stably in all types of tissues and cells of the human body. A well-characterized house-keeping gene is often used as an internal control (IC) for molecular diagnosis (Adeyemi et al., 2017; Mori et al., 2008). The use of human β-globin gene as an endogenous internal control for the EIC-RAA assay is similar to the internal control for Cobas (Roche, Switzerland) and BD onclarity (Becton, Dickinson and Company, USA), HPV testing methods approved by the US Food and Drug Administration (FDA) (Salazar et al., 2019). Unlike exogenously added competitive internal controls, which can only monitor the gene amplification reaction, the endogenous human β-globin can also monitor the sample collection and nucleic acid extraction process. The EIC-RAA can therefore effectively eliminate any false-negative or invalid amplification results, increasing the reliability of the assay. In our study, 3 condyloma acuminatum swabs were negative for human β-globin by qPCR and EIC-RAA. We repeatedly tested the extracted nucleic acid, the results did not change, ruling out errors in the course of the experiment. Then, nucleic acid was extracted from the original sample and tested, and the results remained unchanged, ruling out the possibility of nucleic acid extraction failure. We speculate that manual operation error may have occurred during sample collection. The nucleic acid extraction methods mainly include column extraction, boiling, and magnetic beads, which are a time consuming and complex multi-step process, often requiring skillful personnel (Haukanes and Kvam, 1993; Hirama et al., 2015; Zhang et al., 2019). In contrast, the sample releasing agent allows DNA extraction with just two steps, agent lysis and high-speed centrifugation, performed at room temperature. This provides significant advantages of simplified operation, reduced time and equipment requirement. However, the optimal working condition of the sample releasing agent needs to be determined prior to sample detection, as ingredient in the concentrated solution or nucleic acid releasing agent may inhibit enzymatic activity in the RAA system. Our results demonstrated a 100% consistency rate between sample releasing agent-treated DNA and multi-step extracted DNA, indicating that the EIC-RAA assay has the potential for field detection of HPV6 and 11 in resource-poor areas. One shortcoming of this study is that the number of condyloma acuminatum swab samples was small, and more samples of different types, such as surgical condyloma acuminatum sample, should be evaluated by the EIC-RAA assay. In addition, the RAA detection instrument currently has only two fluorescent channels, preventing simultaneous detection of HPV6, HPV11 and human β-globin in the same reaction. Other portable thermostat with four fluorescent channels are worthy of testing to expand this assay.
We developed a novel EIC-RAA assay for the detection of HPV6 and HPV11 in this study. With comparable sensitivity, specificity to qPCR, the EIC-RAA method is superior in detection speed and operation simplicity.
Anna He: Conceived and designed the experiments; Performed the experiments; Wrote the paper. Cheng Fang, Yue Ming, He Tan, Mengyi Zhang and Yaxing Hu: Contributed reagents, materials, analysis tools or data. Ruiqing Zhang, Jingyi Li, Mingzhu Nie, and Fengyu Li: Analyzed and interpreted the data. Xuejun Ma, Xinxin Shen and Xiuge Rong: Conceived and designed the experiments; Wrote the paper.
This work was supported by IVDC [2019HYDQNJJ03], the key R & D projects in zibo city [2020kj100011].
Data included in article/supp. material/referenced in article.
The authors declare no conflict of interest.
No additional information is available for this paper. |
|
PMC9647409 | Wei Sheng Yap,Guillaume Thibault | A new combinatorial megaplasmid library assembly method designed to screen for minimal pathways by using SCRaMbLE | 26-10-2022 | Human proteins expressed in yeast are common to enhance protein production while the expression of functional human pathways remain challenging. Here, we propose a simple and economical high-throughput gene assembly method to create a yeast megaplasmid library from human cDNA to screen for minimal human functional pathways. We introduced artificial promoters followed by symmetric loxP sites into the megaplasmids using Golden Gate assembly coupled with streptavidin-bead-based purification. The isolated high molecular weight, randomly assembled cDNA megaplasmid library may be useful for high-throughput directed evolution experiments and may be adapted for use in other model organisms. | A new combinatorial megaplasmid library assembly method designed to screen for minimal pathways by using SCRaMbLE
Human proteins expressed in yeast are common to enhance protein production while the expression of functional human pathways remain challenging. Here, we propose a simple and economical high-throughput gene assembly method to create a yeast megaplasmid library from human cDNA to screen for minimal human functional pathways. We introduced artificial promoters followed by symmetric loxP sites into the megaplasmids using Golden Gate assembly coupled with streptavidin-bead-based purification. The isolated high molecular weight, randomly assembled cDNA megaplasmid library may be useful for high-throughput directed evolution experiments and may be adapted for use in other model organisms.
The laboratory model organism S. cerevisiae strain S288C has been used since the early nineteenth century and has become a powerful genetic tool ( Mortimer and Johnston, 1986 ). Despite being highly divergent, human and yeast share many conserved genes and pathways. Thus, it is unsurprising that many scientific breakthroughs are directly attributed to S . cerevisiae , including Nobel Prizes related to cell cycle regulators, telomeres, and autophagy. Subsequently, these discoveries were expanded in human and other higher eukaryotes. However, the complexity of the human genetic architecture remains a roadblock to new discoveries, including parallel pathways and a large proportion of unknown gene functions. Humanized yeast, a concept whereby human genes are expressed in yeast for further characterizations, has been widely practised in different degrees to study the complexity of the human system ( Kachroo et al., 2015 , Laurent et al., 2020 , Laurent et al., 2016 ). Many studies have expressed human orthologs to assess gene complementarity. Recently, with the advent of synthetic biology tools and pathway engineering, the entire human glycosylation pathway was engineered in yeast, signifying the modifiability and adaptability of yeast in modelling higher-order eukaryotes ( Hamilton et al., 2003 , Hamilton et al., 2006 ). However, this is possible only when a pathway is relatively well characterized ( Laurent et al., 2016 ). To this end, we propose a method combining bottom-up and top-down approaches to screen for the minimal human genes required to form a functional pathway in yeast. We developed a method inspired by the recently ongoing synthetic yeast genome Sc2.0 project. The Sc2.0 project started with a bottom-up approach to build a synthetic chromosome containing only the nonredundant genome. In addition, a symmetric loxP site (loxPsym) was incorporated into the non-essential genomic regions between each coding sequence. This loxPsym feature is designed for the downstream directed evolution experiment via synthetic chromosome rearrangement and modification by loxPsym-mediated evolution (SCRaMbLE). SCRaMbLE is a powerful tool in the Sc2.0 project for many applications such as the creation of minimal cells and strains with desired phenotypes. Many studies that utilized SCRaMbLE have had high impact especially in the biotechnology field such as the generation of alkali and ethanol tolerance strains ( Ma et al., 2019 , Luo et al., 2018 ). SCRaMbLE was also used to expedite the identification of synthetic lethal interactions ( Wang et al., 2020 ). While the Sc2.0 project is indeed impressive, it is very costly due to the excessive reliance on gene synthesis and the tedious stepwise gene integration in yeast. Instead of building a whole chromosome out of synthesized gene fragments, we propose to apply SCRaMbLE to a yeast expression megaplasmid of randomly assembled human cDNA to identify the members required for a minimal pathway of desired phenotypes. This system is heterologous and therefore prevents the compensatory mechanisms that otherwise would occur in a same-species expression system. The megaplasmid can be maintained separately from the yeast genome and therefore eliminating the need for in vivo gene editing. We expect that upon completion of the megaplasmid library, it can be characterized genetically and phenotypically, and to be subjected to directed evolution to isolate strains with desired phenotypes (Fig. 1A). We developed and optimized a blueprint of the megaplasmid library in which the vector backbone is a yeast centromeric shuttle vector and the inserts consist of the combinatorial human cDNA library (Fig. 1B). To construct the megaplasmid library, we first performed the template-switching reverse transcription for directional first strand cDNA synthesis which allowed the incorporation of the synthetic minimal promoter near the 5’ UTR of the human cDNA (Fig. 1C). The minimal promoter sequence was modified from a previously reported sequence ( Redden and Alper, 2015 ). We validated the promoter activity through the detection of a GFP reporter (Fig. 1D). To generate a representative and full-length double-stranded cDNA library, we amplified the first strand cDNA using low number of PCR cycles to avoid dNTP exhaustion. Next, we removed short cDNA fragments and other impurities by separating the PCR products on 1% agarose gel (Fig. 1E). To assemble the cDNA fragments into megachunks, we used Golden Gate assembly with PaqCI restriction enzyme which minimizes domestication issues seen in other commonly used enzymes that recognize hexanucleotide sequences. We first tested the assembly method using PCR-amplified fragments of a specific size, a 300 bp CYC1 terminator sequence. We amplified the fragments using two different primer pairs in separate reactions so that the sticky ends produced by PaqCI digestion during Golden Gate assembly were four random nucleotides (NNNN) or TCGA overhangs, respectively. The rationale was to assess if the heterogeneity of sticky ends prevents self-circularization and promotes greater concatemerization. However, the same degree of concatemer formation was observed in both reactions, suggesting that the concatemerization ability was insignificantly influenced by the overhang mismatch (Fig. 1F, left panel). Given this insight, we proceeded with Golden Gate assembly of the cDNA library fragments using the “TCGA” overhang design. This approach should also generate the symmetric spacer sequence of the loxPsym site upon restriction digestion and ligation. As expected, high molecular weight concatemers of the cDNA library fragments were observed after Golden Gate assembly (Fig. 1F, right panel). Next, we prepared the vector backbone by linearizing the centromeric plasmid pGT731 by PCR where one of the primers was 5’-end-biotinylated for subsequent immobilization steps (Fig. 1B). The non-biotinylated end of the backbone vector was cleaved by the restriction enzyme XhoI to produce a sticky end compatible to the “TCGA” overhang produced by the PaqCI digestion. Next, we performed Golden Gate assembly of the vector and the loxPsym-engineered cDNA library, followed by pulldown of the assembled megaplasmid using streptavidin magnetic beads. A distinctive band of the immobilized vector at ~5 kbp was observed in the bound fraction of the negative control (without PaqCI and T4 Ligase), whereas the band corresponding to the vector in the Golden Gate assembly reaction smeared and shifted upwards, indicating successful ligation of the heterogenous cDNA library fragments with the vector (Fig. 1G). The assembled and immobilized DNA products were digested with SgrDI to cut near the biotinylated ends and to generate compatible sticky ends. The released DNA products were self-circularized by T4 ligase and separated on gel electrophoresis for extraction of high molecular weight megaplasmids by electroelution. We transformed the megaplasmids into E. coli for propagation. Probably owing to the cryptic sites and the repetitive sequences within the cDNA library, the colonies harbouring the megaplasmids were relatively smaller in size. Using a pair of vector-specific primers, we amplified the cDNA library inserts by colony PCR to validate the ligation product. The gel electrophoresis revealed the presence (colony 2, 3 and 4) or absence (colony 1) of ligated inserts (Fig. 1H). Intriguingly, colony 5 showed neither the empty vector band nor an upshifted band indicating the presence of an insert. We hypothesized that it could be due to the presence of a huge chunk of insert which is beyond the amplification capability of a conventional PCR. Therefore, we extracted the plasmids from colony 1 and 5 to compare their sizes using gel electrophoresis. Indeed, we noticed the presence of a megaplasmid in colony 5. It is noteworthy that the megaplasmid is in circular and possibly supercoiled form, suggesting that the megaplasmid may be well over 50 kbp in size. Historically, DNA shuffling techniques were employed for the evolution of single known protein coding sequences, de novo chimeric protein construction, and peptide aptamer screening ( Stemmer, 1994b , Stemmer, 1994a , Fujishima et al., 2015 ). Here, we applied the DNA shuffling technique to express a pool of human proteins in yeast for screening of minimal pathways. Our work demonstrates a streamlined method to create a randomized combinatorial megaplasmid library. In synthetic biology, creating a synthetic chromosome is technically challenging. The Sc2.0 project created the synthetic chromosomes by replicating the essential sequences from the yeast genome. In our system, replicating the human genome to be used directly in yeast is not feasible due to the incompatibility of the transcriptional machinery. Instead, we designed a synthetic minimal promoter for heterologous expression to minimize genome instability as a result of repetitive sequences. A better approach is to diversify the short promoter sequences, particularly the spacer regions, although it may incur a higher cost from gene synthesis ( Kotopka and Smolke, 2020 ). The heterologous expression of human proteins in yeast may be toxic especially in high concentration. Therefore, the addition of a controllable cis-regulatory transcriptional silencer may be advantageous. The assembly of a megaplasmid is extremely challenging considering the scale of the molecular construction. Golden Gate assembly was chosen for seamless ligation among the cDNA fragments and between the ligated cDNA fragments and the destination vector. The linearized vector was biotinylated at one end and restriction digested on the other to prevent self-circularization and promote concatenation. Moreover, the biotin conjugated to the vector allowed bead-based purification which removed most of the unligated cDNA fragments that interfere with the electrophoretic analysis, facilitating the extraction of the high molecular weight megaplasmid during the next step. In conclusion, this bottom-up assembly method presented a cheap alternative for the construction of combinatorial library of high complexity. Moreover, the complexity of the library can be easily manipulated, for instance, by transforming S. cerevisiae with multiple megaplasmids of different markers. This method can possibly be modified and adapted for use in other model organisms.
Strains. S. cerevisiae wild-type (WT) strain background BY4741 ( his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 ) was used in this study. In Figure 1D, WT strain was transformed with empty plasmid pGT562 or GFP reporter plasmid pGT784, respectively, using standard lithium acetate transformation protocols. Plasmids used in this study. Plasmids used in this study are listed in Table 1. Plasmids were constructed either by restriction or Gibson cloning. The pGT731 plasmid was cloned by inserting an additional SgrDI restriction site into a SacI/XbaI double digested pRS416 vector. The GFP reporter plasmid, pGT784, was cloned by the assembly of the pGT570 backbone and the synthetic minimal promoter insert. The backbone was amplified from pGT570 using WS170 primer pair to remove the promoter, whereas the synthetic minimal promoter insert was amplified from the cDNA library using TATA-TSS_F and TSS_R primer pair. Flow cytometry. Yeast cells expressing either the synthetic-minimal-promoter-driven GFP or the empty vector control were grown to mid-log phase at 30ºC. The fluorescence intensity was measured using the LSRFortessa X-20 (BD) flow cytometer by exciting at 488 nm and collecting through a 505 nm longpass filter and a 530/30 bandpass filter. A median readout from 10,000 cells was acquired via the software FACSDiVA v 8.0. Data was analysed with the software FlowJo 10.8.0. Reported GFP fluorescence levels were normalized to autofluorescence from the empty vector control. Human cDNA yeast expression library preparation. Total human mRNA was extracted from RPE-1 cells using NEB Magnetic mRNA Isolation Kit (NEB S1550S). Extracted mRNA was used for cDNA synthesis by template-switching reverse transcription (NEB M0466). The oligo (dT)18 primer, WS220, used during the reverse transcription contains an adaptor region of half a loxPsym site and a PaqCI restriction site whereas the template-switching primer, TSO contains the full sequence of the synthetic minimal promoter (Fig. 1C). The cDNA products were treated with Exonuclease I to remove residual primers, and then amplified using WS219 and WS220 primers which added half of a loxPsym site and a PaqCI restriction site to the 5’ end of the products. The amplified cDNA library was column purified (Biobasic BS664) and ran through the size exclusion spin-column (Takara Bio Inc. 636079) for size fractionation. Megaplasmid assembly and purification. To prepare the vector backbone, pGT731 was amplified with WS165 primer pair and column purified. The purified backbone was then incubated with XhoI, DpnI, and CIP for restriction digestion and dephosphorylation. The digested backbone was column-purified again. The backbone was combined with the cDNA library at a 1:5 mass ratio for Golden Gate assembly with PaqCI restriction enzyme (NEB R0745). The Golden Gate assembly reaction was set up according to the manufacturer’s protocol. After the assembly, the linear DNA plasmid was captured by the streptavidin-conjugated magnetic beads (Invitrogen 11206D). The beads were reconstituted with the binding and washing (BW) buffer (5 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, 1 M NaCl) prior to DNA coupling. The DNA-bound beads were washed twice with BW buffer before reconstituting in buffer R (Thermo Scientific BR5). After equilibrating in buffer R, the coupled plasmid DNA was released by SgrDI restriction digestion (Thermo Scientific ER2031). The released plasmid DNA was self-circularized by the addition of T4 DNA ligase (NEB M0202) and ATP (NEB P0756) according to the manufacturer’s protocol. The circular plasmid DNA was run on a 0.6% agarose gel electrophoresis for extraction of high molecular weight megaplasmids by electroelution (G-Biosciences 786-001). Bacterial transformation by electroporation. The electroeluted megaplasmids were transformed into Stable Competent E. coli (NEB C3040H) by electroporation. In brief, 1 µl of the megaplasmids was added to 50 µl of electrocompetent cells and mixed on ice. The mixture was transferred to a pre-chilled 0.2 cm gap width electroporation cuvette (Bio-Rad 1652086) and pulsed at 2.5 kV with the Bio-Rad Gene Pulser Xcell Electroporation System. The electroporated cells were resuspended in 1 ml of outgrowth medium and recovered for 1 h at 37ºC. Cells were then plated on ampicillin selection plates and incubated at room temperature until the appearance of colonies. Colony PCR. The E. coli transformants of relatively smaller colony size were picked and transferred to individual PCR reaction mix. The vector-specific universal primers of T3 and T7 promoters were used for the PCR amplification. The PCR products were then run on a 0.8% agarose gel electrophoresis and visualized afterwards. The presence of a ~170 bp band indicates an empty vector, whereas the presence of an upshifted band indicates the presence of ligated inserts. Megaplasmid extraction from E. coli. E. coli transformants were each inoculated in a 3 ml selective medium and grown at room temperature for 2 days. Alkaline lysis was performed using the reagents provided by the plasmid DNA miniprep kit (Biobasic BS614) according to the manufacturer’s protocol. After the neutralization step, the lysate was clarified by centrifugation and the removal of the pellet fraction. The clarified lysate containing the plasmid DNA was precipitated by the addition of an equal volume of pure isopropanol and centrifugation. The pellet was washed twice and air-dried. Finally, the pellet was resuspended gently in elution buffer using a cut pipette tip. Table 1. Plasmids used in this study Table 2. Oligonucleotide Primers used in this study |
||
PMC9647416 | Manrong Wu,Kunal Rai | Demystifying extrachromosomal DNA circles: Categories, biogenesis, and cancer therapeutics | 26-10-2022 | eccDNAs,eccDNA classification,Biogenesis,Physiological response,Cancer therapeutics | Since the advent of sequencing technologies in the 1990s, researchers have focused on the association between aberrations in chromosomal DNA and disease. However, not all forms of the DNA are linear and chromosomal. Extrachromosomal circular DNAs (eccDNAs) are double-stranded, closed-circled DNA constructs free from the chromosome that reside in the nuclei. Although widely overlooked, the eccDNAs have recently gained attention for their potential roles in physiological response, intratumoral heterogeneity and cancer therapeutics. In this review, we summarize the history, classifications, biogenesis, and highlight recent progresses on the emerging topic of eccDNAs and comment on their potential application as biomarkers in clinical settings. | Demystifying extrachromosomal DNA circles: Categories, biogenesis, and cancer therapeutics
Since the advent of sequencing technologies in the 1990s, researchers have focused on the association between aberrations in chromosomal DNA and disease. However, not all forms of the DNA are linear and chromosomal. Extrachromosomal circular DNAs (eccDNAs) are double-stranded, closed-circled DNA constructs free from the chromosome that reside in the nuclei. Although widely overlooked, the eccDNAs have recently gained attention for their potential roles in physiological response, intratumoral heterogeneity and cancer therapeutics. In this review, we summarize the history, classifications, biogenesis, and highlight recent progresses on the emerging topic of eccDNAs and comment on their potential application as biomarkers in clinical settings.
The ring/disk shaped atypical chromosomes in Crepis tectorum[1] and maize[2] have long been reported in the early 1930’s. Unlike ordinary rod-like chromosomes, these chromosomal derived extrachromosomal circular DNAs (eccDNAs) were modified in their organization and number in different cells, yielding in a typical variegation. Several decades later, Yasuo Hotta and Alix Bassel discovered various sizes of eccDNAs in isolated wheat nuclei and boar sperm by sedimentation analysis and electron microscopy[3], which provided one of the early evidences to support Stahl’s idea that DNA might be circularized in higher organisms[4]. Contemporaneously, Cox et al. encountered various number of small double chromatin bodies neighboring intact chromosomes while karyotyping embryonic tumors and bronchial carcinoma tumor[5], which enhanced the credibility of the existence of small double fragments reported precedingly in a primary lesion of medulloblastoma[6]. Although it was unrealizable to trace their origin to any chromosomes, Cox et al. suggested the acentric ring-like chromatin bodies were not caused by random chromosomal fragmentation[5]. The foremost discovery of eccDNAs was later recapitulated in several other organisms such as the fly[7], hamster[8], mice[9], yeast[10], roundworms[11], pigeons[12], and Arabidopsis[13], suggesting that eccDNAs are prevalent and likely influence cellular processes in eukaryotic cells. While eccDNAs have been identified in both normal and cancer cells, variations in their size distribution [14], [15] and frequencies have been reported[16], [17]. In general, circular structures as large as 104-107 base pairs that carry oncogenes were rarely detected in normal tissues, whereas smaller structures such as small polydispersed circular DNAs (spcDNA) were found in both normal and cancer cells[16], albeit their amount was lower in healthy individuals. Previous attempts to identify and resolve the complex eccDNA elements were constrained by low throughput methods[18]. For example, while both electron microscopy and metaphase 4′,6-diamidino-2-phenylindole (DAPI) could recognize the intactness of these molecules, they were compensated for their low sensitivity and their inability to resolve molecular architecture. Recent advances in next-generation sequencing technologies and third-generation sequencing platforms have revolutionized the way researchers decipher the complex genetic landscape of eccDNAs. Using whole genome sequencing (WGS), cytogenetic and semi-automated image analyses, Turner et al. identified eccDNAs in approximately half of the 17 different cancer types tested, howbeit their frequencies varied based on the tumor types[19]. Similarly, more recently, Kumar et al. used chromatin accessibility assays (i.e. ATAC-seq) to discover thousands of eccDNAs in various cancer types, which were further validated by inverse PCR and metaphase fluorescence in situ hybridization (FISH)[20]. EccDNAs can also be identified by Circle-seq purification and enrichment[21] coupled with long-read sequencing technologies such as long-read Nanopore and single-molecule real-time sequencing (SMRT-seq)[22]. Further, another method called CRISPR-CATCH, which does not require DNA amplification to purify targeted megabase-sized eccDNAs, was invented to overcome limitations of Circle-seq (such as the need for intact DNA circles and the fragility of large eccDNAs)[23]. More recently, a third-generation sequencing technology-based method was developed to enable detection of eccDNAs at a single-cell whole-genome level[24]. Together, the use of parallel paired-end next-generation sequencing by these studies suggest that the architecture of eccDNAs are significantly more complex than previously considered.
EccDNAs are categorized into multiple groups depending on their size and sequence[25], the four[26] that are commonly found in cancer are described below and their method of detection has been summarized in Table 1.
Despite their discovery and isolation based on their buoyancy in alkaline solutions[27], these heterogeneously sized DNA species ranging from 0.2 to 2 µm remained unnamed until 1972, when their name was coined by Smith et al.[28]. Since then, various spcDNA of more than 0.5 µm or 1.5 kb have been identified using mica-press-adsorption for electron microscopy[29]. Although spcDNA have long been suggested to be associated with genomic instability, their varied size and sequence content implies potentially different mechanisms of generation[17]. For example, preferential formation of spcDNA from Alu-rich regions in HeLa cells may be attributed to the juxtaposition of poly(A) sequence at the 5′ and the 3′ end of the Alu element[30]. Alternatively, spcDNA circularization could also result from the recombination mechanism. Homologous intrachromosomal recombination was proposed by Jones and Potter to explain the 9 bp direct repeats that occur in spcDNA[31]. Repeated circularization of multiple recombination events in the Vα and Jα regions of T-cell receptor α-chain could also explain the high copy number of spcDNA[32]. Whilst spcDNA are generated through hitherto unknown processes, mechanisms may exist that enable elements to loop out of the chromosomes and promote the joining of flanking DNA by illegitimate recombination[30].
In 1995, Nosek et al. reported the discovery of inverted terminal repeats that were made up of tandemly repeating units in yeast type 2 linear mitochondrial genomes (mtDNA)[33]. Taking Candida parapsilosis as an example, the terminus is comprised of a 738 bp repeating unit, with a 5′ single-stranded extension of about 110 nucleotides that is accessible to the enzymes. Later, Tomaska et al. used electrophoresis and electron microscope to reveal that the super-twisted circular conformation of these extragenomic molecules was derived from mitochondrial telomere repeats[34]. Although minicircular structures in mitochondrial DNA have been reported in several phylogenetically distinct species[39], [40], [41], [42], [43], their precise mechanism of generation and functions remains unexplored. While t-circles could be generated either by intramolecular recombination within the telomeric array or via telomeric loop extrusion, it was believed that t-circles maintain telomeric arrays of linear DNA through recombination. As DNA ends need to be protected against nucleolytic attacks and improper DNA metabolism, Tomaska et al. suggested that the various types of t-elements represent alternative strategies to tackle these obstacles[44]. For example, t-circles are believed to maintain telomeres integrity through alternative lengthening of telomeres (ALT) in 15 % of telomerase-negative cancers[45], [46], [47], thereby providing an alternative strategy for aberrant upregulation of cancer cell activity.
A new form of eccDNA – microDNA was first found in 2012 in various tissues and cell lines[35]. Spanning 200 to 400 base pair long, these microDNA are enriched in 5′ UTR, exons, and CpG islands, and have a short region of micro-homology at the beginning and the end of the circles, suggesting the likelihood of microdeletions from the source genomic loci[35]. Unlike spcDNA, which typically span a few kilobase pairs and seem to originate from repetitive regions[7], [8], [9], [48], [49], [50], microDNA originate from non-repetitive sequences, preferentially from high gene density areas. In 2015, Dillon et al. revealed microDNA were originated from DNA breaks or replication slippage following mismatch repair and loop excision[51]. The same study upon profiling microDNA from chicken DT40 cell lines lacking various crucial DNA repair proteins involved in non-homologous end joining (NHEJ), homologous recombination (HR), and microhomology-mediated end-joining (MMEJ), observed that microDNA were produced by all mutant strains, confirming that no single DNA repair pathways was responsible for generating microDNA[51]. In addition to these three potential mechanisms, Dillon et al. also pointed out these extra copies of genomic regions could alter cellular functions by protein titration and abnormal short RNAs production. More recently, using synthesized microDNA that resembled known microDNA regions, Paulsen et al. showed that microDNA express functional small regulatory RNA that are subsequently processed into mature microRNA (miRNA) and repressed endogenous targets[52]. Since microDNA that carry miRNA genes are functional, and miRNAs in turn are indispensable to animal development, cell differentiation and homeostasis[53], and tumor progression[54], it is likely that microDNA are key regulators of biological processes.
In 2017, Turner et al. introduced yet another new type of eccDNA – ecDNA, a mega base pair amplified circular DNA that is visible in optical microscopy[19]. Accordingly, multiple image-based analysis tools were developed to identify ecDNA from DAPI-stained metaphases[19], [38]. Unlike other eccDNA, ecDNA are almost never found in normal cells, but are large enough to carry driver oncogenes[19]. A subsequent study using AmpliconArchitect found that oncogenes amplified on ecDNA had higher transcripts compared to when the same genes were not amplified on ecDNA, even after normalization of their copy numbers, suggesting alterations to the genetic structure, such as enhanced chromatin accessibility[55]. Although the underlying mechanisms of ecDNA biogenesis are not yet fully elucidated, four models have been proposed, including breakage-fusion-bridge cycle, translocation-excision-deletion-amplification, episome, and chromothripsis. Details of these four models are further examined in the next section.
This concept was first introduced in 1951 by McClintock while studying the mechanisms responsible for mutable loci in maize[56]. The breakage-fusion-bridge cycle is initiated when newly broken ends of chromosomes at a meiotic mitosis cause the fusion between sister chromatids (Fig. 1A), resulting in a bridge configuration followed by separation of centromeres of the dicentric chromatid[56]. Because the break could occur anywhere between the two centromeres and the cycle continues in successive mitoses during development, breakage-fusion-bridge cycle results in extensive DNA ladder-like focal amplifications and large deletions, and potentially the formation of double minutes (DM)[57], [58], [59], a term often used in early studies to describe extrachromosomal structures.
Translocation and amplification are two important cytogenetic categories associated with tumor etiology[60]. The chromosomal translocations studied in various cancers suggested two routes to activate oncogenes: the activation of a proto-oncogene juxtaposed to a T-cell receptor gene or an immunoglobulin protein, and the creation of a fusion gene by the breaks of two coding regions[61]. DNA amplification is a frequent genetic abnormality in tumors, which are manifested as DM and homogenously staining regions (HSR) as cytogenetic hallmarks[62]. In some cases, translocation could concert with amplification to promote tumorigenesis. In 1996, Barr et al. employed FISH, RT-PCR and Southern blot to show the amplification of PAX3-FKHR or PAX7-FKHR fusion genes in 20 % of fusion-positive alveolar rhabdomyosarcomas[60], substantiating a sequential process through which oncogenes were activated. Another study unraveled the mechanism of non-syntenic co-amplification of MYC and ATBF1 in a neuroblastoma cell line, which involved multiple double-stranded breaks accompanied by a reciprocal t(8;16) translocation and deletion near the breakpoints[63]. In line with other findings[64], [65], extra replication or loop formation could result in a DM configuration (Fig. 1B).
The concept of episome was first introduced in 1987 when Carroll et al. found a subclone of T5 transformant gave rise to a CAD episome containing donated CAD genes[66]. Gel electrophoresis showed these extrachromosomal molecules were 250 to 300 kilobase pairs in size and were covalently closed. Like DM, these circular elements contain a functional origin of DNA replication and can replicate autonomously. To explore the episomal formation mechanism, Carroll et al. grew CAD episome containing T5 cells under nonselective conditions and found that the loss of episome was correlated to the loss of donated CAD genes, suggesting the formation of episome by corresponding chromosomal region deletion[67]. Two mechanisms were speculated by the authors, re-replication model and recombination across the looped replication domains[67]. While the prediction of the first model implies a rereplicated chromatid strand forming a “loop” structure, the latter model involves recombination of donated chromosomal sequences of sufficient size bearing origins of replication[68]. Another study on MYC carrying a DM has shown the amplified region was deleted at 8q24 in 68 % of the cases, which favored the episome model[69]. Interestingly, breakage across replication bubbles at stalled forks could also result in ecDNA formation (Fig. 1C)[70], [71]. Consequently, episomal amplification could result in palindromic amplicons[71], or tyrosine kinase activation by the fusion between NUP214 and ABL1, promoting the pathogenesis of T-cell acute lymphoblastic leukemia (T-ALL)[72].
A remarkable phenomenon whereby extensive genomic rearrangements occur in a single catastrophic event was termed as chromothripsis by Stephens et al. in 2011[73]. Although the prevailing cancer evolution dogma indicated gradual acquisition of driver mutations, which resulted in increasing malignancy[74], somatic mutation outbursts might be a one-time event which promotes cancer development[73]. Importantly, chromothripsis could facilitate ecDNA generation (Fig. 1D)[73], [75], [76]. For example, the MYC containing DM was found to be generated by the shattering of chromosome 8 in a small cell lung cancer cell line[73]. While the mechanism for chromothripsis is unknown, a recent study showed that the process was dependent on poly(ADP-ribose) polymerases (PARP) and DNA-dependent protein kinase (DNA-PKcs)[77]. An error resulting in chromosome mis-segregation or intact chromatin bridge during the interphase could also pulverize the chromosomes[78], causing the scars in genome and DNA rearrangements, usually resulting in DNA circularization.
The diverse molecular and physiological functions of EccDNA summarized in Fig. 2 are described in detail this section.
Antibiotic resistance (AbR) genes reside on mobile genetic elements (MGEs) such as plasmids, integrative and conjugative elements (ICEs), and various transposons[79]. Like plasmid DNA, eccDNAs carry critical genes that offer selective advantages in varying selective pressures (reviewed below in the section Adaptation of eccDNAs under therapeutic response). Since MGEs are mobile through various mechanisms, it is likely that eccDNAs are locomotive[80]. One aspect of eccDNA motility is its elimination by micronucleation[81], [82], [83], which was exemplified by the spontaneous extrusion of supernumerary MYCN amplified eccDNA into micronuclei[82]. It was shown that in a small percentage of cells, hybridization signals distributed peculiarly in clusters, adhered to nuclear membrane, and aggregated in nuclear protrusions[82], [84], suggesting a spontaneous elimination process. The idea of an eccDNA hub was confirmed in a later study[85] where the authors labelled MYC ecDNAs with TetR-eGFP/TetR-eGFP(A206K) in COLO320-DM cells. Interestingly, treatment with 500 nM of a bromodomain and extraterminal (BET) protein inhibitor, JQ1, dispersed eccDNA hubs in COLO320-DM cells but did not alter the signal distribution of MYC in COLO320-HSR cells, implying the involvement of BET in hub maintenance[85]. Subsequently, Yi et al. established a CRISPR-based tracking technique which utilized sequences covering eccDNA-specific breakpoints to uncover disjointed eccDNA inheritance pattern during mitosis[86]. The authors found that the fluorescent signal was diluted during cellular cytoplasmic division and was reestablished once the two daughter cells entered the interphase; thereby providing direct visual evidence of the spatiotemporal dynamic feature of eccDNA[86]. Another aspect of eccDNA mobility is exemplified by the HIV-1 DNA integration into and disintegration from the host genome. 1-long terminal repeat (1-LTR) and 2-LTR circles are the two types of episomal HIV-1 DNAs that are particularly found in acutely infected cells such as the effector memory CD4+ T cells[87]. Although these elements contain sequences for viral replication, their functions remain unknown and were considered as by-products of the reverse transcription. A recent study suggested that 2-LTR circles could serve as reservoirs for proviral integration due to their palindromic junctions being recognized by integrase. Interestingly, the cleavage was specific and could be improved by the integrase cofactor LEDGF/p75[88]. It is likely that 2-LTR could serve as a main source of substrate for integration when its number surpasses that of linear DNA. For example, in the presence of a HIV integrase inhibitor raltegravir[88]. In a recent study, clustered regularly interspaced short palindromic repeats/Cas9 (CRISPR/Cas9) was used to excised HIV proviral DNA in NL4-3/Luc-transduced 293 T cells[89]. Upon ablation, circular DNA with full-length LTRs formed via intermolecular, and sense-sense joining could be detected for up to 14 days. These concatemers upregulated integrase and p24 production upon pTat and pRev cotransfection, suggesting that they could be transcriptionally active[89]. Like retroviruses, retrotransposons also involve reverse transcribed eccDNAs as part of their lifecycles[90]; therefore, eccDNAs could potentially characterize the reservoir of active transposons[13], [36], [90], [91]. Interestingly, transposon display revealed integration of ONSEN transposons into the genome in Arabidopsis under heat stress and drugs, suggesting eccDNAs may contribute to genome evolution[90].
Due to the absence of centromeres, eccDNA are subjected to loss during nuclear envelope break down. However, studies revealed that eccDNA were tethered to chromosomes, which enabled acentric eccDNA to be efficiently passed onto daughter cells during mitosis[92], [93], [94]. Interestingly, eccDNA were associated with the periphery of prometaphase chromosome rosettes and were localized far away from the spindle poles, suggesting their dependence on antipolar forces[93]. Although the mechanisms of eccDNA-chromosomal adherence remains unclear, a study on the interaction between the origin of plasmid replication (oriP) and the viral protein EBNA-1 in Epstein-Barr virus (EBV) offered an unique insight[95]. The trans-activator protein EBNA-1 interacted with oriP and was thought to facilitate the anchorage of viral genomes on cellular chromosomes[96]. Analogously, Baiker et al. have shown the interaction between the origin of replication in simian virus 40 (SV40) genome, which attached to scaffold/matrix attachment region (S/MAR), and the chromosome scaffold, providing a mechanistic explanation of episomal stability and retention[97]. Although the idea that eccDNA might interact with chromosomes or with each other is not new[17], [98], it was speculative and no concrete evidence was revealed. Recently, a model proposed ecDNA functioned as mobile regulatory elements that promoted the activity of chromosomal genes[99]. In this study, the authors performed Hi-C and RNAPII-associated ChIA-PET analysis on multiple GBM-patient-derived neurospheres and found ecDNA broadly contact the whole genome. The enrichment of trans-chromosomal interaction frequencies (nTIF) compared to the average genome-wide nTIF at 50 kb resolution was still significant after adjusting for the copy number. Furthermore, by comparing Histone H3 Lysine 27 acetyl (H3K27ac) peaks detected in the interacting loci on the ecDNAs, their chromosomal partners, and the genome-wide regions not contacting with ecDNAs, the authors showed ecDNA-chromosome interactions were associated with transcriptionally active sites[99]. Multiple lines of evidence have also corroborated the ecDNA-chromosomal interactome including: the significantly higher RNA transcription of chromosomal genes contacting ecDNAs, multi-color FISH validation, the higher interaction frequencies mediated from ecDNA by comparing adjusted nTIF of subsampled regions on ecDNA and chromosomes with matched H3K27ac fold enrichment, the higher RNA expression level of genes connecting with ecDNA but with comparable RNAPII binding enrichment than those without connections. Although the dynamics of eccDNA diffusion and the stability of trans-interaction remain unknown, ecDNA could potentially act as mobile enhancers (Fig. 3) that greatly expand the transcriptional plasticity of a cell population[100].
Although oncogene bearing eccDNAs could facilitate gene overexpression by merely increasing their copy number[19], the amount of DNA template was not the only factor that contributed to gene transcription[55], [101], [102], [103]. Wu et al. integrated ATAC-seq profile with WGS data and found that the ATAC-seq signal was significantly higher in circular amplicons even after normalization of the DNA copy number[55]. In addition to the less compacted nucleosomal organization, co-amplification of the proximal enhancer and hijacking of the ectopic enhancer into highly rearranged MYCN amplified ecDNA have also been reported[101]. This finding was in line with an earlier study that found significant co-amplification of non-coding DNA beyond amplified oncogenes on ecDNA across several tumor types[102]. Interestingly, a recent study showed guide RNAs targeting an intergenic region near MYC- amplified ecDNAs significantly impaired cell growth[104]. Together these studies highlight new mechanisms by which eccDNA contribute to cancer progression.
A recent study suggested that eccDNAs could function as potential innate immunostimulants in a manner which was dependent on the circular structure but not the underlying sequence[105]. The authors generated bone marrow-derived dendritic cells and bone marrow-derived macrophages, and compared their immune response (i.e. production of IFNα, IFNβ, IL-6, and TNF) between linear genomic DNA, eccDNA, and poly(dG:dC). Surprisingly, all aforementioned cytokines were significantly generated by eccDNA compared to linear DNA at varying concentrations. Similarly, these cytokines were significantly upregulated by eccDNA compared to poly(dG:dC) at lower concentrations, implicating the potency of the circular DNA on immune response. Furthermore, generation of linearized eccDNA by introducing one nick per circular DNA revealed that these linearized eccDNA behaved like linear DNAs and failed to activate cytokines; thus supporting the strong immunostimulant activity of eccDNA[105] and their potential induction of primary B cell and T helper type 2 responses[106]. Circularization of viral DNA by NHEJ pathway, on the other hand, has been suggested to alleviate the apoptotic effect in retroviral infected cells, albeit without excluding indirect mechanisms for suppression[107]. Although unintegrated viral DNA expression could not downregulate human leukocyte antigen complex on resting CD4+ T cells, it sensitized infected cells to be targeted and killed by functional cytotoxic T cells[108]. Moreover, episomal HIV-1 DNA has been shown to be transcriptionally active, and could generate functional viral proteins such as Tat and Nef[109], [110], [111], [112], which may contribute to the CC-chemokine production in infected macrophages and the recruitment of lymphocytes[113].
The capability of eccDNAs to dynamically regulate themselves under environmental pressures is well known. In 1979, Kaufman et al. observed the unstably amplified dihydrofolate reductase gene (DHFR) on DM in higher concentrations of methotrexate[114]. As DHFR converts dihydrofolic acid to tetrahydrofolic acid, which is indispensable for synthesizing purines and pyrimidines, the presence of DHFR-carrying DM may partly explain methotrexate resistance. Similarly, later studies also identified an association between amplified DHFR and small circular DM in various cell types[104], [115], [116], [117]. Surprisingly, DHFR was altered spontaneously and exhibited a 270-fold reduction in binding affinity for methotrexate[115], which further contributed to the inhibitor insensitivity by reducing the enzymatic activity. When methotrexate was absent, the growth rate of methotrexate-resistant murine S-180 cell was inversely correlated to the copy number of DHFR carrying DM[118], implying that the cells lacking DHFR amplification had a growth advantage when the drug was removed, and were likely selected during uneven segregation. Interestingly, the resistant S-180 line lost most of its DM after continued selection in methotrexate-containing medium and acquired DHFR genes on a few chromosomes[118]. This phenomenon was also reported in other studies[119], [120], [121], which could be revealed by Giemsa-banding. In addition to the association between DHFR carrying eccDNA and antifolate resistance, the loss of oncogene bearing eccDNA was also correlated with the substantial reduction of tumorigenicity in several human tumor cell lines[81]. Von Hoff et al. showed that the treatment of a low concentration of hydroxyurea on HL60, COLO 320, NB4, and SF188 cells promoted the loss of MYC amplified DM, which is involved in the entrapment of DM within the micronuclei[81]. In a clinical trial conducted in 2001, researchers investigated whether a low dosage of hydroxyurea could downgrade DMs in 16 patients with advanced ovarian carcinomas[122]. Results revealed that 45 % of the patients showed more than 50 % reduction of the number of spreads with DM containing tumor cells, while one patient demonstrated 52 % reduction of the c-myc copy number after hydroxyurea treatment[122]. Similar mechanism of eccDNA extrusion was also proposed in another study[123], where 2 Gy fractions up to a total radiation dose of 28 Gy resulted in the reduction of MDR1 and MYCC bearing eccDNA via entrapment in micronuclei in multidrug-resistant lines. Other chemotherapeutic agents have also been reported to regulate gene amplifications on eccDNAs. For example, lower levels of mitoxantrone induced ABCG2 amplification via DM in the SF295 glioblastoma cell line[124]. Gemcitabine at a 7500X lower concentration of hydroxyurea effectively reduced DM in an ovarian cancer cell line by incorporating the amplicons and γ-H2AX signals into micronuclei[125]. The sensitivity of cisplatin induced apoptosis could be reversed by introducing antisense oligonucleotides targeting against MDM2 mRNA in human glioblastoma cells[126]. As the development of resistance to chemotherapeutic drugs is considered as one of the reasons for the failure of cancer treatment, and both DM and HSR were shown as cytogenetic manifestations in cancer cells, studying the role of eccDNAs in chemotherapeutic drug response may offer mechanistic insights into acquired resistance. Targeted cancer therapies that interfere with tumor cell growth by interacting with specific molecules have shown some promising results as evident from the high response rates demonstrated by patients on these regimens. However, the prolonged benefit of majority of these therapies are limited by the eventual development of resistance of the tumor cells[127], [128], [129]. Nathanson et al. have found that resistance to erlotinib, an epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor (TKI), in glioblastoma was conferred by the reversible loss of eccDNAs containing the active oncogenic variant EGFRvIII, which bestowed an optimal cellular state for growth and survival[130]. This finding was later recapitulated in another study[104]. Interestingly, EGFR+ HSRs were observed throughout the entire stages of naïve, drug resistant, and drug retraction, whereas EGFR+ eccDNAs were only completely lost in erlotinib-resistant GBM cells, suggesting an adaptive route by which tumors can evade targeted therapy. Similarly, Song et al. recently showed that BRAF amplification also causes challenges to targeted therapy dosages[131]. The authors developed a melanoma model of dual MAPK inhibitor (MAPKi; specifically, vemurafenib [BRAFi] and selumetinib [MEKi]) resistance that bore BRAFV600 amplifications either through DM or HSR. They found that drug-resistant plasticity was coupled with focal amplifications, and that inconstant drug dosage prevented the switch from DM to HSR. Moreover, a different form of cell death, ferroptosis, occurred during BRAF amplification mediated MAPKi resistance, extending the resistant mechanisms beyond cellular dedifferentiation[131].
EccDNA dynamics has also been suggested to promote plant evolution. Koo et al. reported that eccDNA based EPSPS amplification was associated with rapid glyphosate resistance in the crop weed Amaranthus palmeri through adaptive evolution[132]. Interestingly, a sexual transmission study that crossed a female A. palmeri lacking eccDNA and a male A. palmeri carrying eccDNA showed positive signals associated with mitotic metaphase chromosomes in the descendants, indicating the successful transmission of herbicide resistance to the offspring via eccDNA.
Recently, Hull et al. demonstrated the transcription of tandem CUP1 copies stimulated CUP1 encoding eccDNA in yeast that were aged under environmental exposure to copper[133], a process triggered by factors such as Sae2, Mre11 and Mus81 that are involved in DNA repair. In addition to facilitating adaptive evolution, eccDNAs accumulation in the nuclei may also promote ageing and tumorigenesis. Current hypothesis suggests the transport of damaged DNA via eccDNA out from the nucleus into the cytosol, where a cell-autonomous nucleic acids (NA) degradation machinery is triggered to keep NA below the immunostimulatory threshold[134]. However, during ageing, the increase of dysfunctional nuclear pore complexes (NPCs) may result in the accretion of eccDNA in the nucleus[135]. The consequent accumulation of DNA damage in turn promotes cellular senescence and apoptosis[136], further indicating a strong link between eccDNAs and ageing.
Circular RNAs (circRNAs), recently found to be amply supplied and stable in exosomes[137], were suggested to mediate intercellular crosstalk in the tumor microenvironment by involving cancer and stromal cells[138]. For example, hepatocellular carcinoma (HCC) cells with high metastatic capability could transfer their metastatic potentiality to other HCC cells with low or no metastatic capability by secreting exosomes with circPTGR1[139], exosomal circ-CCAC1 in cholangiocarcinoma was transmitted to endothelial cells, which promoted angiogenesis by downregulating junctional proteins[140]. Similarly, could eccDNA trapped micronuclei represent an entity facilitating intercellular network? Despite extracellular micronuclei with DM being reported in some studies[83], [141], [142], which suggested micronuclei could be expelled from the cell and serve as a repertoire of DNA elements, the impact on intercellular genetic communication remains unclear. Interestingly, the content of micronuclei could be shuttled into multivesicular bodies via direct contact[143]. Nevertheless, mitochondrial circular genome could not only be transferred through direct cell–cell contact[144], [145], but also via circulating extracellular vesicles[146], suggesting circular DNAs as potent paracrine/endocrine signaling factors. However, as studies on eccDNAs as communicators between cells are limited and still speculative, additional are needed to determine this potential role of eccDNAs.
Episomal HIV-1 DNAs were found in patients with advanced central nervous system damage[147], and patients on antiretroviral therapy (ART)[148], [149], [150]. Although the roles of these episomes are still debatable, a study comparing the envelop sequences in episomal and proviral genomes before viral rebound upon treatment interruption with those in emergent viral RNA[150] suggested that episomal HIV-1 could fuel viremia rebound. Moreover, episomal HIV-1 genomes could be used as a marker to monitor ART, due to its lability in vivo, and given that traditional methods were not sensitive in viral reservoirs detection[151]. A recent study using four paired primary and metastatic tissues of high grade serous ovarian cancer (HGSOC) highlighted the association between DNMT1circle10302690-10302961 downregulation and HGSOC metastasis[152], suggesting certain eccDNA element could be considered as a prognostic marker. Liquid biopsy, which includes cell-free DNA (cfDNA), circulating tumor cells and exosomes, has made great progress in recent years due to its non-invasiveness and informativeness[153]. EccDNAs are a type of cfDNA in the circulating system and evidence of their role in disease association and progress surveillance suggests their potential to be harnessed as biomarkers. For example, a recent study comparing eccDNAs from the plasma of 6 lung adenocarcinoma (LUAD) and 10 healthy individuals reported of a higher frequency of nine top ranked eccDNAs in LUAD samples when compared to the healthy group. Interestingly, the study also found that DOCK1, PPIC, TBC1D16, and RP11-370A5.1 were uniquely encoded in eccDNAs in LUAD group[154]. Similarly, another study showed that the cell-free microDNA present in tumor lung tissue specimens were longer than those in paired normal lung samples; moreover, serum and plasma samples collected prior to surgery were enriched with longer microDNA compared with that obtained from the same patients following surgical tumor resection[15]. Interestingly, the formation of eccDNAs was found to be dependent on the lineage of cancer[51]. Although unique eccDNAs were able to be identified in some diseases, characteristics such as high GC content, repetitiveness, and low quantity in plasma[154] may hamper their application in clinical setting by increasing the difficulty in primer design and temperature control. While common eccDNA detection methods involves RCA, which is an efficient isothermal DNA amplification procedure, the synthesis cost is high, and primers are still in need. Nevertheless, a recent study using a label-free fluorescent biosensor to detect circRNA provides an ultrasensitive alternative to identify eccDNA[155]. While circRNA was reported to degrade completely within 15 s in 25 % serum[156], the average half-life of fetal eccDNA in the maternal blood was found to be 29.7 min[157]. The half-lives of extracellular microRNAs (ex-miRNAs), on the other hand, varied between ex-miRNA entities and the species[158], [159]. It is important to note that while the stability of ex-miRNAs were measured in cell culture [158], [159], fetal eccDNA kinetics were determined from blood samples collected from pregnant women before delivery and at multiple time points postpartum[157]. Serial time point collection of blood was important as it correctly reflects the half-lives of eccDNA in the biological system. While long non-coding RNAs (lncRNAs) have drawn attention as molecular biomarker for cancer prognosis[160], their rate of decay has also only been measured in in vitro models[161], [162]. While the half-lives of circRNA, eccDNA, ex-miRNA, lncRNA are yet to be directly compared from genome-wide analyses, based on eccDNAs’ size distribution, sequence, resistance to exonuclease or ribonuclease, and their stability compared to RNA, growing evidence indicates that eccDNAs could serve as potent biomarkers for disease surveillance.
To date, three eccDNA databases have been introduced (Fig. 4)[163], [164], [165], which compiled eccDNAs from different resources and are focused on diverse aspects of eccDNAs. While CircleBase, the first integrated platform for eccDNA functional interpretation, comprises 601,036 eccDNAs collected from 13 papers[163], both eccDNAdb[164] and eccDB[165] profiled eccDNAs based on different library types and computational methods. To characterize potential eccDNA function, six modules comprising of various functional databases were used, and the diffusion algorithm PageRank was used to prioritize genes interacting with eccDNAs based on three of the modules[163]. While eccDNAdb focused on the prognostic value of eccDNA genes[164], eccDB complemented CircleBase database by incorporating eccDNA interchromosomal interactions and evolutionary relationships through multi-species sequence comparisons[165].
In this review, we summarize the current state of understanding of eccDNAs pertaining to their discovery, prevalence across multiple species and cancer types, classification and associated formation mechanisms, and physiological characteristics. Although the knowledge that eccDNAs contribute to cancer progression has been known for decades, gaps in our knowledge of eccDNAs remain and continue to be one of the intractable challenges faced by cancer researchers[166]. Although the overall frequency of DM in primary cancer has been reported to be 1.4 % based on the Mitelman database[167], [168], a recent study suggests its frequency is far greater than previously inferred[19]. 1. Future research on eccDNA integration and dissociation is needed. Since sequence and structure could contribute to the selective fragility of the genome[169], [170], what are specific features of the boundaries. Are there specific chromatin organization or underlying DNA sequences? 2. While the prominent mechanism of how eccDNAs promote evolution of cancer cells relies on competitive advantage offered by uneven segregation of oncogene amplified eccDNAs during mitosis, the detailed molecular mechanisms are only now beginning to emerge and will require detailed mechanistic studies[84], [104], [171], [172], [173]. An important contributor to this puzzle could be that eccDNA can behave like enhancer elements and may traverse the nucleus to enable global chromosomal contacts thus introducing transcriptional plasticity in a cell population. Colocalization of eccDNAs which form a hub and recruit RNA polymerase not only provides a plausible mechanism of high oncogene transcriptional rate and intratumoral heterogeneity, but also offers promising avenues for the eccDNA hub directed cancer therapies. 3. Another important outstanding question is if DMs replicate independently. While some studies suggested DMs replicate synchronously with chromosomal in S phase[174], [175], others suggested they were derived de novo from HSR fragmentation[176]. Nevertheless, the separation of DM’s sister elements during G1 phase prevents the formation of quadruple chromatids[177], and likely alleviate their numerical heterogeneity between cells which results from their anomalous segregation during cell cycle. The mechanism of separation is unclear because the chromatin fiber organization of DMs is currently unknown[177]. Furthermore, DMs attach to nucleolar matter near the end of chromosome arm during metaphase[92], but what force drives scattered DMs inside the cytoplasm in prophase to relocate to the chromosome ends in metaphase? 4. Finally, aside from the elucidation of the eccDNA replication mechanism during cellular division, more studies on eccDNAs’ roles in clonal dynamics, their targetable vulnerabilities, and their half-lives are needed to fully realize their translational value.
Manrong Wu: Writing – original draft. Kunal Rai: Supervision, Writing – review & editing.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. |
|
PMC9647421 | Jiayu Zhang,Jun Chen | Circular mRNAs: More stable RNAs lead to more persistent protein expression | 08-11-2022 | circular mRNA,cmRNA,linear mRNA,internal ribosome entry site,IRES,cytokines,LNP | Circular mRNAs: More stable RNAs lead to more persistent protein expression
Yang et al. established a novel type of circular mRNA called cmRNA. cmRNA contains an echovirus 29-derived internal ribosomal entry point element to promote ribosome binding and a newly designed spacer to enhance translation. Their findings suggest that this type of circular mRNA can mediate strong and persistent expression of various types of proteins. Furthermore, in addition to delivery via lipid nanoparticle (LNP), cmRNAs were discovered suitable for direct intratumoral administration. By giving mice cmRNAs encoding mixed cytokines directly intratumorally, they achieved successful modulation of intratumoral and systemic antitumor immune responses and enhanced anti-programmed cell death protein 1 (PD-1) antibody-induced tumor suppression. Natural circular RNA is an emerging member of the RNA family that has gained importance in research due to its novel functional role in cell physiology and disease progression. Since RNA degradation starts at the tail end and there are no tail ends after cyclization, circular RNAs stay longer than linear ones. Various techniques for in vitro circularization of RNAs have been established after decades of research, including enzymatic ligation via T4 ligases, chemical strategies for the ligation of 5' and 3' ends, and a ribozyme technique that creates an intramolecular covalent connection. The greatest potential of circular mRNAs is its ability to meet therapeutic needs that mRNA cannot, which enables long-acting protein replacement or protein overexpression RNA therapies. In addition, circular mRNAs are more promising because they are less immunogenic and more readily available. In this work, the authors designed a novel cmRNA. It carries echovirus-derived internal ribosome entry site (IRES) elements and spacers screened by point mutations, resulting in sustained and efficient protein expression. Additionally, the authors found that high-performance liquid chromatography (HPLC) purification of cmRNA was necessary to eliminate immunogenicity and promote protein expression. Subsequently, with the help of LNP, sustained protein expression was obtained by administration of cmRNA in vivo and in vitro. To directly alter the intratumoral and systemic anticancer immune response, cmRNA encoding mixed cytokines was injected into tumors, increasing the tumor-suppressive effect of anti- PD-1 antibodies in a mouse model. Moreover, CaCl2 and KCl were found to be essential for the intratumoral delivery of cmRNA. By adjusting the dosage of both, the intratumoral delivery exhibited better anticancer effects and led to substantial infiltration and activation of CD4+ and CD8+ T cells. One of the most exciting aspects of this manuscript is that naked cmRNA can be delivered directly into tumors and express the relevant proteins by intratumoral injection. In mice tumor models, 6 h after intratumoral injection of cmRNA with luciferase coding sequences, significant luminescence was detected in all transplanted tumors, indicating that cmRNA was delivered and expressed in tumor tissue. This finding implies that cmRNA can be taken up by cancer cells without vectors, eliminating the need for delivery vector development and concerns about material safety. Nevertheless, cmRNA may not perform well in targeted delivery. Effective luciferase activity was found in the liver and the injection site 24 h after intramuscular injection, indicating that LNPs primarily transported cmRNA to the liver. The LNP technology, although proven safe and effective, is limited in its hepatic targeting, hindering the application of cmRNA drugs. Therefore, new delivery strategies are urgently demanded for extrahepatic targeting of cmRNA medicines. Overall, this work demonstrates the future trends of RNA drugs. First, circular mRNA with better stability and immunogenicity may replace linear mRNA; Second, naked circular mRNA may be injected into tumor tissues through image guidance such as computed tomography (CT) to improve the ability of immune cells to attack cancer cells. |
||
PMC9647423 | Lianxiang Luo,Xinming Chen | Exploring the potential of eRNAs in cancer immunotherapy | 08-11-2022 | Exploring the potential of eRNAs in cancer immunotherapy
The recent success of RNA therapy for non-neoplastic diseases has fueled the development of various RNA-based cancer immunotherapies. Enhancer RNA (eRNA), short noncoding RNA (50–200 nucleotides) transcribed from the enhancer region, can affect anticancer drug resistance, which may be associated with one or more related tumor signaling pathways. This research has led to explorations of novel eRNA-based cancer treatments. Bioinformatics analyses have revealed functional eRNA molecular subtypes in lung adenocarcinoma. However, the genome-wide eRNA landscape and its relevance to the immune microenvironment in hepatocellular carcinoma (HCC) remain undefined. In this issue, Bu et al. share a novel exploration of eRNA transcriptional regulation in HCC, which compelled us to examine how eRNAs regulate genes in cancer and the value of eRNAs in cancer immunotherapy.
Human eRNAs are present in clusters associated with genes that are collaborating partners in protein complexes. This “functional cooperation” may be common during development to dictate transcription factor expression. Within clustered enhancers, redundancy and hierarchy can coexist in various combinations to guarantee functional cooperation on target genes. More research is needed to evaluate eRNA transcription as an independent criterion for predicting active enhancers and annotating non-transcribed enhancers. RNA surveillance pathways that block the decay of eRNAs can be exploited to detect less stable eRNAs, facilitating the characterization of putative non-transcriptional enhancers. Measuring eRNA levels can identify enhancer features in physiological and pathological conditions. Moreover, given their specificity for cell type and state, eRNAs may provide diagnostic and therapeutic targets. Another important goal is better describing enhancer-promoter loops, preferably at the single-cell level. This will confirm the role of eRNAs in regulating circularization and further delineate the relationship between circularization, eRNAs, and gene transcription. Given the commonality between enhancers and promoters, we must determine their mutual regulation and underlying hierarchy in the loading of transcriptional machinery. A potential approach is systematically deleting promoters and enhancers to study their relationship in three-dimensional genomes.
dCas9-based activators can induce eRNA production, which is positively correlated with mRNA expression downstream of the homologous promoter. For example, KHPS1 activates eRNA through triplex-dependent recruitment of epigenetic regulators and increases the expression of proto-oncogene SPHK1. Nuclear effector YAP1/TEAD4 interacts with ERα-binding enhancer, which shows enhanced binding after E2 stimulation to facilitate E2/ERα target gene induction and E2-induced oncogenic cell growth. eRNA SEELA facilitates histone recognition and oncogene transcription by enhancing interactions between chromatin and histone modifiers. The SEELA-SERINC2 axis also regulates cancer metabolism, potentially influencing leukemia progression.
Many pivotal host genes of Kaposi’s sarcoma-associated herpesvirus (KSHV) latency, including proto-oncogene MYC, are controlled by super enhancers. In KSHV-infected primary effusion lymphoma (PEL), the eRNA-expressing MYC super enhancer is located downstream of MYC. Deletion or inhibition of eRNA expression significantly reduces MYC mRNA in PEL. Similarly, in colon cancer cells, the WNT/β-catenin-AHCTF1-CTCF-eRNA circuit enables the oncogenic super enhancer to promote cancer cell growth by coordinating the trafficking of the active MYC gene in the three-dimensional (3D) nuclear structure. p53 binding enhancer regions also have enhancer activity and interact with neighboring genes in chromosomes to transmit long-distance transcriptional regulation. Léveillé et al. explored p53-regulated enhancers by examining eRNA production. They found that long noncoding RNA (lncRNA) lncRNA activator of enhancer domains (LED) activates strong enhancers and that LED knockdown attenuates p53 function. In human colorectal cancer cells, bromodomain-containing protein 4 (BRD4) is recruited to enhancers cooccupied by mutant p53 and supports the synthesis of enhancer-directed transcripts (eRNAs) in response to chronic immune signals. eRNA directly participates in gene regulation by regulating BRD4’s enhancer interaction and transcriptional function. BRD4 acts as an anti-breast cancer target to promote p63 and GRHL3 expression downstream of FOXO in breast epithelial cells. Interestingly, BRD4-dependent BIRC3 eRNA synthesis confers Helicobacter pylori-mediated apoptosis resistance.
eRNA enhances gene activation via the androgen receptor (AR) drive loop. AR binds enhancer elements and regulates specific enhancer promoter loops, thus activating the AR regulatory network. Kallikrein-related peptidase 3 (KLK3) is an AR regulatory gene that encodes prostate-specific antigen (PSA). KLK3’s upstream enhancer generates a bidirectional enhancer in RNA, KLK3e. Functional antisense eRNAs negatively regulate the antisense ncRNAs of AR-related target gene sites in prostate cancer cells by recruiting DNMT1 on antisense enhancers in gene terminal regions and increasing methylation. Importantly, a chromatin double loop promotes the cis sense to the promoter and the antisense to the gene terminal region. Deleting antisense eRNA impairs adjacent mRNA expression and impedes cancer progression. Antisense eRNA expression correlates with biochemical recurrence and clinical marker PSA level in patient tissues. Hormone therapy leads to enhancer landscape recombination in breast cancer cells. Upstream of oncogene DDIT4, glucocorticoid receptor (GR) binds to four sites that constitute hormone-dependent hyper enhancers. Three GR binding sites are required as hormone-dependent enhancers to differentially promote histone acetylation, transcription frequency, and burst size. By contrast, the fourth site inhibits transcription, but hormone treatment alleviates this inhibition. An estrogen-responsive RNA, P2RY2 eRNA (P2RY2e), participates in breast cancer development. CRISPR-Cas13a-mediated knockdown of P2RY2e inhibits the proliferation, invasion, and migration of bladder cancer cells; it may also weaken the tumor promoting effect of estrogen on bladder cancer.
Uncontrolled enhancer activity with aberrant eRNA expression may promote tumorigenesis. Specific transcriptional eRNAs from active enhancers may serve as therapeutic targets in certain cancers. For instance, Bu et al. found that eRNAs in immune-related cluster 1 are enriched in immune infiltration and may respond to immune checkpoint inhibitors. However, to utilize eRNAs as therapeutic targets in cancer, we need to identify more eRNA classes and mechanisms. One technology, Cap Analysis of Gene Expression, can capture transcriptional eRNAs and identify enhancers with extremely high nucleotide resolution. Thus, clinical treatment prediction models constructed using eRNAs have broad prospects. In general, eRNA plays a crucial role in regulating gene transcription. Additionally, eRNAs associated with hormone receptors play regulatory roles in multiple cancers. Mining the types and functions of eRNAs will contribute to the revolutionary application of eRNAs to cancer immunotherapy. |
|||
PMC9647424 | Qin Jiang,Chunlin Xie,Lingli Chen,Hongli Xiao,Zhilian Xie,Xiaoyan Zhu,Libao Ma,Xianghua Yan | Identification of gut microbes associated with feed efficiency by daily-phase feeding strategy in growing-finishing pigs | 29-09-2022 | Daily-phase feeding,Feed efficiency,Gut microbiota,Nutrient metabolism,Pig | Feed efficiency is one of the most important issues for sustainable pig production. Daily-phase feeding (DPF) is a form of precision feeding that could improve feed efficiency in pigs. Gut microbiota can regulate host nutrient digestion, absorption, and metabolism. However, which key microbes may play a vital role in improving the feed efficiency during DPF remains unclear. In the present study, we used a DPF program compared to a three-phase feeding (TPF) program in growing-finishing pigs to investigate the effects of gut microbiota on feed efficiency. A total of 204 Landrace × Yorkshire pigs (75 d) were randomly assigned into 2 treatments. Each treatment was replicated 8 times with 13 to 15 pigs per replicate pen. Pigs in the TPF group were fed with a commercial feeding program that supplied fixed feed for phases I, II, and III, starting at 81, 101, and 132 d of age, respectively, and pigs in the DPF group were fed a blend of adjacent phase feed from 81 to 155 d at a gradual daily ratio and phase III feed from 155 to 180 d of age. Daily feed intake and body weight were recorded by a computerized device in the feeders. Feces and blood samples were collected from 1 pig per replicate at 155 and 180 d of age. The results showed that the DPF program remarkably improved the feed efficiency at 155 d (P < 0.001) and 180 d of age (P < 0.001), with a significant reduction of the intake of crude protein (P < 0.01), net energy (P < 0.001), crude fiber (P < 0.001), ether extract (P < 0.01), and ash (P < 0.001). The daily-phase feeding program increased the abundance of Prevotella copri (P < 0.05) and Paraprevotella clara (P < 0.05), while it decreased the abundance of Ocilibacter (P < 0.05) at 155 d of age. The results of correlation analysis indicated that the differentially abundant microbiota communities were closely associated with 20 metabolites which enriched amino acid and phenylalanine metabolism. Our results suggest that 2 key microbes may contribute to feed efficiency during daily-phase feeding strategies in pigs. | Identification of gut microbes associated with feed efficiency by daily-phase feeding strategy in growing-finishing pigs
Feed efficiency is one of the most important issues for sustainable pig production. Daily-phase feeding (DPF) is a form of precision feeding that could improve feed efficiency in pigs. Gut microbiota can regulate host nutrient digestion, absorption, and metabolism. However, which key microbes may play a vital role in improving the feed efficiency during DPF remains unclear. In the present study, we used a DPF program compared to a three-phase feeding (TPF) program in growing-finishing pigs to investigate the effects of gut microbiota on feed efficiency. A total of 204 Landrace × Yorkshire pigs (75 d) were randomly assigned into 2 treatments. Each treatment was replicated 8 times with 13 to 15 pigs per replicate pen. Pigs in the TPF group were fed with a commercial feeding program that supplied fixed feed for phases I, II, and III, starting at 81, 101, and 132 d of age, respectively, and pigs in the DPF group were fed a blend of adjacent phase feed from 81 to 155 d at a gradual daily ratio and phase III feed from 155 to 180 d of age. Daily feed intake and body weight were recorded by a computerized device in the feeders. Feces and blood samples were collected from 1 pig per replicate at 155 and 180 d of age. The results showed that the DPF program remarkably improved the feed efficiency at 155 d (P < 0.001) and 180 d of age (P < 0.001), with a significant reduction of the intake of crude protein (P < 0.01), net energy (P < 0.001), crude fiber (P < 0.001), ether extract (P < 0.01), and ash (P < 0.001). The daily-phase feeding program increased the abundance of Prevotella copri (P < 0.05) and Paraprevotella clara (P < 0.05), while it decreased the abundance of Ocilibacter (P < 0.05) at 155 d of age. The results of correlation analysis indicated that the differentially abundant microbiota communities were closely associated with 20 metabolites which enriched amino acid and phenylalanine metabolism. Our results suggest that 2 key microbes may contribute to feed efficiency during daily-phase feeding strategies in pigs.
A stable pork supply plays a vital role in addressing how to feed the world and provide animal protein sources in the coming decades (Lassaletta et al., 2019). However, pork production has large environmental challenges, such as greenhouse gas emissions, feed demand, land pollution, and freshwater use (Rauw et al., 2020; Tilman et al., 2002). Methods of sustainable swine production include genetic modification, reproductive techniques, environmental stressors, and feeding strategies (Ernst and Steibel, 2013; Wu et al., 2020; O'Connor et al., 2010). Feed efficiency has high impact on the sustainable pig industry due to feed costing up to 60% to 75% of total production cost (Godinho et al., 2018; Patience et al., 2015; Wu et al., 2020). Precision feeding is an integrated approach that allows for feed demands with consideration for changes in nutrient requirements of each pig in the herd to optimize performance at minimal feed costs (Pomar et al., 2014), and is also a way to improve feed efficiency. Of note, daily-phase feeding (DPF) is a form of precision feeding that can be widely applied in farms (Andretta et al., 2014). The majority of gut microbiota contribute to nutrient digestion and absorption, cellular growth, and tissue homeostasis in the host, and also play an important role in maintaining gut health in mammals (David et al., 2014; Gentile and Weir 2018; Lynch and Pedersen 2016; Subramanian et al., 2015). The pig genome and gut microbiome have been reported (Chen et al., 2021a, Chen et al., 2021b; Groenen et al., 2012; Xiao et al., 2016), allowing a better understanding of how the interaction between host and gut microbiota affects the feed efficiency of swine (Bergamaschi et al., 2020; McCormack et al., 2017; Wang et al., 2019; Yang et al., 2017). Daily-phase feeding strategies can largely alter feed efficiency during swine production. Meanwhile, most current feeding strategies in China such as three-phase feeding (TPF) or five-phase feeding cannot avoid imbalances between swine nutrient requirements and dietary nutrient supply because daily changes in body weight are ignored (Pomar et al., 2021). Given the close relationship between gut microbiota and host nutritional metabolism in pigs (Gardiner et al., 2020; Patil et al., 2020), we need to investigate which key microbes play vital roles in improving feed efficiency during a DPF program. In this study, we firstly investigated how a DPF strategy affected feed efficiency by analyzing the growth performance, intestinal morphology, and the intestinal functions. We then identified those microbes related to feed efficiency in growing-finishing pigs. Using 16S rDNA gene sequencing at 155 and 180 d of age, we tested the effect of a DPF program in contrast to a TPF program on pig growth performance and on the composition and potential functions of the intestinal microbiome. Finally, we further assessed the plasma metabolome profiles and analyzed the relationship between the potential key microbes and metabolites, which may contribute to the understanding of the interaction of host and microbiota on feed efficiency of swine.
All experiments involving swine were carried out in strict accordance with the Guide for the Care and Use of Laboratory Animals Monitoring Committee of Hubei Province, China, and the protocols were approved by Institutional Animal Care and Use Committee of Huazhong Agricultural University, Wuhan, China, under permit number HZAUSW-2019-020.
In this study, 204 Landrace × Yorkshire pigs were raised in a commercial farm (Wuhan, Hubei, China). All experimental pigs were moved to an environmentally controlled fattening house (randomly assigned to 13–15 pigs in each pen) of the age of 75 d, and 101 and 103 pigs in 8 pens were assigned to the TPF program and DPF program, respectively. Pigs were fed with a commercial nursing feed for 5 d before the experiment. The pigs had ad libitum access to water and feed throughout the trial. Each pig had an electronic chip placed in their ear that enabled them to access to the feeders and the feeding station identified the pig when its head entered the feeder and constantly provided 30 g of feed when the feeder hopper was emptied. The meal size was calculated by recording the amount of food delivered in the short intervals between 2 consecutive visits registered by Electronic Feed Intake Recording Equipment (FIRE, Osborne, USA). Individual feed intake and body weight data were collected by FIRE from 80 to 180 d of age, and used to calculate growth performance indicators, such as average daily feed intake (ADFI), average daily gain (ADG), and feed conversion ratio (FCR). The feeder calibration was checked weekly. At 155 and 180 d, we randomly chose 1 pig per pen to collect fresh fecal and blood samples. Fecal samples were collected during or within less than 30 s after defecation, and blood samples were collected via jugular venipuncture into vacuum tubes and kept frozen in liquid nitrogen for transportation, and then stored at −80 °C until use. At 180 d of age, 1 pig from each pen was sacrificed for further sample collection.
The experiment used 4 commercial feeds supplied by COFCO Corporation. Diets were formulated to meet the nutrient requirements of swine (NRC, 2012) and the nutritional values are shown in Table 1. The schematic design of conventional phase feed and daily mixed feed delivery to the pig pens is shown in Fig. 1A. Compared with the single feed silo that supplied 1 feed directly to 1 pig pen in the TPF program, the DPF program applied mixed feed with a gradual daily change of 2 adjacent phase feeds from 2 feed silos to 2 pig pens. The TPF program was supplied with fixed feed for phase I, phase II, and phase III which started at 81, 101, and 132 d of age, respectively. We designed the DPF program using adjacent phase feed based on an empirical model which predicts pig growth and its nutrient requirements (Hauschild et al., 2010). Detailed programs were performed with a blend of nursing feed and phase I feed mixed with a varied composition from 81 to 90 d of age, and a mix of phase I feed and phase II feed from 91 to 115 d of age, and a mix of phase II feed and phase III feed from 116 to 155 d of age according to Table S1. The simulated changes in nutrient levels are shown in Fig. S1. The blend formula was changed every day at 08:00 as the unconsumed diets were recycled to non-experimental pens and the new mixed diets were supplied to the feeder from 81 to 155 d of age in DPF, and from 156 to 180 d of age, pigs in DPF were fed with phase III feed that was same as the TPF program. The experimental design and sample collection time points are illustrated in Fig. 1B.
To evaluate the intestinal morphology, duodenal, jejunal, and ileal tissues fixed in 4% paraformaldehyde were embedded in paraffin and sliced, 5 μm paraffin sections were dewaxed with xylene, hydrated, and then stained with hematoxylin and eosin (H&E). For each sample, 8 intact villi-crypt units were selected for morphology observation using a light microscope coupled with Image J software (Rasband, NIH, USA). Villus height (VH, the height from the tip of villus to the villus-crypt junction) and crypt depth (CD, the depth of invagination between adjacent villi) were measured. Villus height/crypt depth was calculated.
Blood samples were placed on ice immediately after collection and centrifuged at 3,000 × g for 15 min. Serum samples were separated and stored in plastic tubes frozen at liquid nitrogen until analysis. The concentrations of serum total protein, blood urea nitrogen (BUN), alkaline phosphatase (ALP), albumin, glutamic-pyruvic transaminase (ALT), glutamic-oxalacetic transaminase (AST), creatinine, and triglyceride were measured by an autoanalyzer (Abbott Alcyon 300, Abbott Diagnostics, Lake Forest, IL) and Abbott reagents (Table 2). Serum endotoxin, diamine oxidase, D-lactic acid concentration were analyzed, using commercial kits (Nanjing Jiancheng Bioengineering Institute of China, Nanjing, China), following the protocols described by the manufacturers. Approximately 0.1 g of jejunal tissue was mixed with 1 mL phosphate-buffered saline (PBS) and homogenized after milling at 12,000 × g for 10 min at 4 °C. Cytokine concentrations of supernatants were measured with enzyme-linked immunosorbent assay (ELISA) kits (Nanjing Jiancheng Bioengineering Institute of China, Nanjing, China). The activities of amylase, maltase, lactase, sucrase, lipase, and pepsin of the duodenum, jejunum, and pancreas were determined by colorimetry using assay kits (Nanjing Jiancheng Bioengineering Institute of China, Nanjing, China).
The total genomic DNA of fecal bacteria was extracted using a commercial kit (QIAGEN QIAamp PowerFecalDNA Kit, Germany). The integrity of the DNA was assessed by agarose gel electrophoresis. Genomic DNA was used as a template for PCR amplification as previously described (Xu et al., 2021). Universal primers 338F and 806R were used for PCR amplification of the V3–V4 hypervariable regions of 16S rRNA genes (338F, 5′-ACTCCTACGGGAGGCAGCA-3′; 806R, 5′-GGACTACHVGGGTWTCTAAT-3′). The generated DNA pool was then sequenced on the Illumina HiSeq system with the sequencing strategy PE 300. The sequencing data were analyzed using the Quantitative Insights Into Microbial Ecology (QIIME2) software package. All 16S rDNA sequencing data are publicly available in the National Center for Biotechnology Information (NCBI, PRJNA816606). Then paired end reads with overlap were merged to tags. The high-quality paired-end reads were combined with tags based on overlaps with FLASH. The tags were clustered to amplicon sequence variant (ASV) by scripts of software Dada2. After that, the ASV unique representative sequences were obtained. Chimeras were filtered out by UCHIME (v4.2.40); ASV representative sequences were taxonomically classified using Ribosomal Database Project (RDP) Classifier v.2.2 trained on the database. The alpha diversity and principal coordinate analyses (PCoA) were calculated by Mothur (v1.31.2) with the corresponding rarefaction curve being drawn by software R (v4.2.0). The statistical significance of gut microbiota among different groups was assessed by permutational multivariate analysis of variance (PERMANOVA; 9,999 permutations, P < 0.001) in R. PCoA and PERMANOVA were based on the data matrix of the Bray–Curtis distances. Differential enrichment of bacterial features between the TPF and DPF groups was determined using linear discriminant analysis (LDA) effect size (LEfSe), with the all-against-all multiclass analysis, P < 0.05, and a logarithmic LDA threshold of 4.0. Functional analysis of gut microbiota was predicted by Phylogenetic Investigation of Communities by Reconstruction of Unobserved States 2 (PICRUSt2) (Douglas et al., 2020).
Blood plasma samples were extracted to obtain the supernatants. An equivalent volume was aliquoted from each sample, mixed to prepare the QC sample, and dried in a vacuum concentrator. A high-resolution tandem mass spectrometer Xevo G2 XS QTOF (Waters, UK) was used to detect metabolites eluted from the column. The Q-TOF was operated in both positive and negative ion modes. For positive ion mode, the capillary and sampling cone voltages were set at 3.0 kV and 40.0 V, respectively. For negative ion mode, the capillary and sampling cone voltages were set at 2.0 kV and 40.0 V, respectively. The mass spectrometry data were gained in Centroid MSE mode. The TOF mass range was from 50 to 1,200 Da and the scan time was 0.2 s. For the MS/MS detection, all precursors were fragmented using 20 to 40 eV, and the scan time was 0.2 s. During the acquisition, the LE signal was gained every 3 s to calibrate the mass accuracy. In order to test the stability of the LC-MS during the whole acquisition, a quality control sample (pool of all samples) was gained after every 10 samples. Peak extraction was mainly achieved through the commercial software Progenesis QI (version 2.2), including peak alignment, peak extraction, normalization, deconvolution, and compound identification. Based on QC sample information, local polynomial regression fitting signal correction (Quality control-based robust LOESS signal correction, QC-RSC) was performed on the real sample signal. The obtained data were applied for principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA). R2 and Q2 were used to evaluate the quality of PCA and OPLS-DA models. Differential metabolites were identified with variable importance projection (VIP) > 1.0 and P < 0.05.
Total RNA was extracted from the jejunum of pigs. The relative expression of genes associated with anti-oxidation was measured by quantitative polymerase chain reaction (qPCR). Oligonucleotide primers used to amplify target genes are shown in Table S2. The qPCR program consisted of an initial denaturation at 95 °C (10 min) followed by 40 cycles of denaturation at 95 °C for 15 s, 60 °C for 30 s, 72 °C for 30 s, annealing at 60 °C for 30 s and extension at 72 °C for 30 s. The 2−ΔΔCT method with GAPDH used as the reference gene was used to calculate the relative expression.
Experimental data were analyzed by unpaired two-tailed t-test or Mann–Whitney U test with GraphPad 9.0 software (GraphPad Software Inc., San Diego, USA). The detailed descriptions of the statistical methods are shown in each the legends of each figure. Significance was set as ∗ P < 0.05, ∗∗ P < 0.01, and ∗∗∗ P < 0.001. Results are presented as mean ± standard error of the mean (SEM).
Growth performance from 81 d of age to the end of feed shift time (155 d of age) and the end of the experiment (180 d of age) was not influenced by the treatment (P > 0.05) with regard to ADG and ADFI (Fig. 2A and B). However, FCR was significantly lower in DPF than that in TPF with a remarkable decrease from 2.58 to 2.43 (P < 0.001) at 155 d of age, and the significant differences lasted to 180 d of age with the ratio of 2.73 to 2.63 (P < 0.001) (Fig. 2C). In addition, the DPF program reduced the intake of crude protein (P < 0.01) (Fig. 2D), lysine (P < 0.01) (Fig. 2E), crude fiber (P < 0.001) (Fig. 2F), ash (P < 0.01) (Fig. 2G), ether extract (P < 0.01) (Fig. 2H), net energy (P < 0.01) (Fig. 2I), and total phosphorus (P < 0.05) (Fig. 2J) at 155 and 180 d of age. Available phosphorus intake was significantly reduced at 180 d of age (P < 0.05) (Fig. 2K), but not at 155 d of age (P = 0.0578).
Jejunual and ileal villus height to crypt depth ratio were significantly enhanced in DPF compared to TPF (P < 0.05), while no significant changes were observed in the duodenum (Fig. 3A and B). Compared to TPF, DPF showed a slight increase in amylase, lactase, sucrase and ALP in the duodenum and jejunum, while a decreasing trend in pepsin was detected in the pancreas (Table 3). The production of pro-inflammatory cytokines including tumor necrosis factor-alpha (TNF-α), interleukin-1-beta (IL-1β), and interleukin-6 (IL-6) were measured in the jejunum, with no significant changes observed (Fig. S2A). By measuring the intestinal permeability and anti-oxidative status, we found that pigs in DPF had lower plasma diamine oxidase concentration (P < 0.05) than pigs in TPF (Fig. 3E), whereas the endotoxin and D-lactic acid concentrations were similar between the 2 programs (Fig. 3C–D). The activities of total antioxidant capacity (T-AOC) and total superoxide dismutase (T-SOD) of the jejunum were significantly higher in DPF than the TPF program (P < 0.05) (Fig. 3F and G), and there was an increasing trend in glutathione peroxidase (GSH-PX) (Fig. 3I) in DPF compared to TPF, while there was no difference between DPF and TPF in malondialdehyde (Fig. 3H). The mRNA expression results also showed that SOD1, SOD3 and NOQ1 (P < 0.01) were significantly increased and KEAP1 was decreased (P < 0.05) in DPF compared to the TPF program (Fig. S2B).
Daily-phase feeding had no significant effect on the alpha diversity and richness of the gut microbiota compared to TPF at both 155 and 180 d of age (Fig. 4A–C). PCoA based on Bray–Curtis distances revealed that the gut microbiota of pigs was altered under the different feeding patterns (Fig. 4D). The gut microbiota between TPF and DPF differed significantly at 155 d of age, and it was worth noting that TPF and DPF no longer differed significantly at 180 d of age, as shown by the comparison of distances (Fig. 4E). At the phylum level, Firmicutes, Bacteroidetes, Spirochaetes, and Proteobacteria predominated in both the TPF and DPF programs at 155 and 180 d of age. In comparison to TPF, the profiling of microbial phyla in DPF was characterized by a high proportion of Bacteroidetes and a low proportion of Proteobacteria (Fig. 5A), and the Firmicutes to Bacteroidetes ratio was significantly lower in DPF than the TPF program at 155 d of age (Fig. 5B). At the family level, the 3 most abundant bacterial families both at 155 and 180 d of age primarily consisted of Prevotellaceae, Lactobacillaceae, and Lachnospinaceae. The average abundance of Prevotellaceae of DPF increased to 35.1%, while it was 22.9% in the TPF program (Fig. 5C). At the genus level, 20 taxa accounted for about 95% of the total assigned sequences across systems at 155 d of age (Fig. 5D). Prevotella (30.2%), Lactobacillus (11.7%), Treponema (3.6%), and Lachnoclostridium (2.9%) were the 4 most abundant genera in both programs. Clostridium (3.4%) was notably more abundant in TPF than in DPF (1.8%), while Paraprevotella (2.5%) was more abundant in DPF than in TPF (1.0%). At the species level, pig gut in DPF harbored more Prevotella copri DSM18205 (P < 0.05), and Paraprevotella clara YIT11840 (P < 0.05) but fewer Oscillibacter valericigenes (P < 0.05) in DPF at 155 d of age than in the TPF program (Fig. 6A–E). We determined the correlation between signature gut microbes of the feeding strategy and feed conversion ratio using Spearman's rank correlation coefficient and significance test. The results showed that Oscillibacter (P < 0.05) was positively correlated with feed conversion ratio, and at species level, O. valericigenes was significantly related to FCR (Fig. 6F and G). Thus, we concluded that DPF decreases the relative abundance of Oscillibacter, while increasing the abundance of Paraprevotella and Prevotella compared with TPF.
To better understand the functional roles of the microbiome, we used PICRUSt2 to investigate the functional profiles of the gut bacterial community (Fig. 7). The results demonstrated that the relative abundances of the genes involved in “Methane metabolism”, “Carbon metabolism”, “Ribosome”, and “Biosynthesis of cofactors” pathways were significantly changed in the DPF program. A total of 9 predicted function genes involved in “cell motility” and “signal transduction” were differentially represented between the 2 programs at 155 d of age. As such, DPF pigs had a higher relative abundance of the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway genes involved in “Fructose and mannose metabolism”, “Biosynthesis of cofactors”, “Pyrimidine metabolism”, compared to TPF pigs (P < 0.05). In contrast, the predicted pathways “Pyruvate metabolism”, “Propanoate metabolism”, “Flagellar assembly” and “Phosphotransferase system” were lower in DPF pigs, compared to TPF pigs (P < 0.05). The altered “Digestive system” and “Carbon metabolism” functional pathway provides strong evidence that the DPF program regulates carbohydrate metabolism and amino acid metabolism functions through gut microbiota.
A total of 2,515 plasma metabolite features were detected by the UPLC-QTOF/MS. The PCA of the metabolites between TPF and DPF programs at 155 and 180 d of age were similar to that of the fecal microbiota (Fig. 8A). However, we did not detect the significant relationships between plasma metabolites and the FCR values at the significance threshold of P < 0.05. Metabolites with VIP > 1 and bacterial genera significantly affected by feeding programs were used for the Pearson's correlation analysis. We found that 24 metabolites were significantly deferentially expressed between the 2 programs at 155 d of age, with 16 over-expressed in the TPF program and 8 over-expressed in the DPF program (Fig. 8B). Among these 24 metabolites, 13 metabolites had a tendency to positively correlate with the abundance of Prevotella copri DSM18205, and P. clara YIT11840 (P < 0.05), and one metabolite showed a tendency to positively correlate with O. valericigenes (Fig. 8C). The TPF program enriched pathways related to vitamin B6 metabolism and purine metabolism, and DPF enriched pathways related to amino acid metabolism, including phenylalanine, tyrosine and tryptophan biosynthesis and phenylalanine metabolism (Fig. 8D and E).
We performed this study to better understand microbial differences in the pig gut microbiome between TPF and DPF programs related to feed efficiency. A daily-phase feeding program could have beneficial impacts on feed efficiency, with improved intestinal morphology, intestinal permeability, and the anti-oxidative ability of the intestine, as well as increased abundance of potentially beneficial microbes. Integrative analysis showed that DPF alters the microbiota and that P. clara and Prevotella copri were 2 key microbes related to high levels of phenylalanine metabolites in plasma that may improve feed efficiency. Pigs were fed with 3 or 5 feeding phases, which meant nutrients could be under- or over-supplied during the whole feeding period (Pomar and Remus, 2019). Daily-phase feeding is a practical solution for addressing this imbalance problem by adjusting the whole herd daily and providing nutritional requirements daily, respectively (Andretta et al., 2014). By measuring feed intake, body weight, and carcass composition of each pig, DFP has proven to be more precise in terms of meeting the nutrient requirements of the herd compared with TPF, excretes less nitrogen and phosphorus into the environment, and saves costs. Some studies did not observe that FCR could be reduced in DPF, which may be due to insufficient amino acid supplies (Andretta et al., 2014; Remus et al., 2019). In our study, we found that the DPF program could significantly reduce the FCR and save nutrition compared to the TPF program during the daily mixed feed feeding phases; FCR and these effects persisted even after the DPF group was fed the same feed as the TPF program. The efficiency of nitrogen and phosphorus utilization, however, was not measured. Feed efficiency is a complex phenotype that captures how effectively feed is turned into food products for human beings. A genome-wide association scan was conducted to identify the genetic markers and genes associated with feed efficiency (Do et al., 2013), and the intestinal microbiota of different breeds with feed efficiency traits (Bergamaschi et al., 2020; Do et al., 2014), and microbes that may be influenced by host genetics (Camarinha-Silva et al., 2017) were identified. The gut microbiota of pigs with extremely high and low RFI under the same feeding conditions were examined to find potential gut bacteria associated with feed efficiency (McCormack et al., 2017; Tan et al., 2017; Yang et al., 2017). Other studies focused on influencing feed efficiency via various feeding strategies, such as reducing meal frequency (Yan et al., 2021), dietary protein restriction feeding (Fan et al., 2017), and time-restricted feeding (Zeb et al., 2020), were also asubjected to comparative analysis of the gut microbiota. The gut microbes from fecal microbiota transplantation (FMT) (Siegerstetter et al., 2018) and dietary supplementation with feed additives that could improve feed efficiency were also investigated (Li et al., 2020). Here, we compared the gut microbiota between TPF and DPF programs that differed in feed efficiency, to identify the key microbes that could promote feed efficiency. The gut microbiota of DPF differed the most from those of TPF, whereas the difference was not detectable after the 2 groups were fed the same diet for 25 d. This is consistent with the theory that a long-term dietary pattern may be the most suitable factor for influencing the gut microbiota (David et al., 2014). The gut microbiota of the 2 feeding programs showed this effect, demonstrating the significance of feeding strategy and diet in determining feed efficiency in growing-finishing pigs. Firmicutes and Bacteroidetes, the predominant phyla in the gut microbiome in most mammals, have been closely correlated with energy partitioning and feed efficiency in pigs (Bergamaschi et al., 2020). Phylum Bacteroidetes has been shown to be more abundant in pigs with high feed efficiency or lean pigs compared to those with lower feed efficiency or obese pigs (Jami et al., 2014). While in this study, no significant difference in the abundance of these 2 taxa was observed between TPF and DPF groups, Firmicutes and Bacteroidetes predominated in the fecal microbiome of all pigs, and the relative abundance of Bacteroidetes in the fecal microbiome of DPF pigs with high feed efficiency was slightly higher than that of pigs in the TPF group. Although microbial diversity showed no difference between TPF and DPF, we further identified the potential key gut microbes related to FCR. Our analysis reveals that Paraprevotella, Prevotella and Ocilibacter are correlated with the FCR. Prevotella and P. clara belong to the Prevotellaceae. Prevotella, a core genus of pigs, produces enzymes that can degrade complex dietary polysaccharides, then improve fiber digestibility and feed efficiency (Chen et al., 2021b; Amat et al., 2020). This could explain why Prevotella provided more energy for the host and contributed to feed efficiency. Yang et al. found that Prevotella might be a keystone microbe to increasing host feed intake (Yang et al., 2017). P. clara is one of the predominant cellulolytic bacterial species that is closely related to the production of short-chain fatty acids (Gao et al., 2018). The metabolic pathways of aromatic amino acids (e.g., phenylalanine metabolism), which are related to the synthesis of indole propionic acid, have also been associated with the species from P. clara. Gut microbiota homeostasis influences intestinal metabolism (Lee et al., 2014). We hypothesized that DPF-induced alterations in microbiota composition might be followed by changes in the abundance of various metabolites in plasma. In this study, a comprehensive analysis of the plasma metabolome of TPF and DPF pigs was conducted. Most of the differentially enriched metabolites between TPF and DPF pigs were accounted for as amino acid metabolites. Functional analysis showed that the differential metabolites were involved in phenylalanine, tyrosine and tryptophan biosynthesis and phenylalanine metabolism. In animal models, phenylalanine could help to improve host metabolism by enhancing intestinal barrier function, showing anti-inflammatory properties and strengthening immune function (Leblhuber et al., 2015; Xu et al., 2020). In this study, the 2 main microbes were highly positively related to the increase in amino acid catabolites. Particularly, the metabolite of trans-cinnamic acid, a gut microbial metabolite produced by the deamination of phenylalanine, is known to exhibit a wide range of biological activities such as antioxidant and anti-inflammatory activities (Foti et al., 2004). These results showed that the altered phenylalanine profile in DPF plasma might contribute to feed efficiency in pigs.
In summary, we revealed that a DPF program could improve feed efficiency in growing-finishing pigs via the development of villi, permeability, and the anti-oxidative ability of the intestine, and by altering the abundance of Paraprevotella, Prevotella and Ocilibacter, thus affecting the plasma metabolites in phenylalanine metabolism. Together, the combined results of the present study may aid future research in identifying the 2 key microbes that promote feed efficiency in growing-finishing pigs.
Qin Jiang, Conceptualization, Methodology, Validation, Formal analysis, Investigation, Visualization, Funding acquisition, Writing - Original Draft; Chunlin Xie, Investigation, Validation; Lingli Chen, Investigation, Formal analysis, Visualization; Hongli Xiao, Investigation, Validation; Zhilian Xie, Investigation, Validation; Xiaoyan Zhu, Investigation, Validation; Libao Ma, Supervision, Resources; Xianghua Yan, Conceptualization, Supervision, Project administration, Funding acquisition.
We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, and there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the content of this paper. |
|
PMC9647426 | Carolina Schinke,Alexandra M. Poos,Michael Bauer,Lukas John,Sarah Johnson,Shayu Deshpande,Luis Carrillo,Daisy Alapat,Leo Rasche,Sharmilan Thanendrarajan,Maurizio Zangari,Samer Al Hadidi,Frits van Rhee,Faith Davies,Marc S. Raab,Gareth Morgan,Niels Weinhold | Characterizing the role of the immune microenvironment in multiple myeloma progression at a single-cell level | 20-08-2022 | Visual Abstract Image 1 | Characterizing the role of the immune microenvironment in multiple myeloma progression at a single-cell level
Multiple myeloma (MM) is a malignant disease of plasma cells (PCs) that reside within the bone marrow (BM). The disease transitions from the precursor stages, monoclonal gammopathy of undetermined significance (MGUS) and smoldering MM (SMM), to clinically aggressive disease. Although the outcomes have improved, the disease remains largely incurable once progression has occurred., However, at early premalignant disease stages, treatment is not routinely administered because not all cases progress, and the clinical course is essentially benign. In an effort to define more aggressive variants of these precursor conditions, in which disease intervention strategies would be justified, recent studies have focused on identifying prognostic markers that predict progression to MM. Confirmed markers include clinical parameters such as increased BM content of aberrant PCs,, abnormal light chain ratio in the peripheral blood (PB), evidence of circulating PCs, and the presence of immunoparesis (abnormal-to-normal PC ratio and reduced noninvolved immunoglobulin)., Furthermore, the genomic alterations translocation t(4;14), a gain of chromosome 1q, TP53 mutations, and MYC translocations are strongly associated with progression., Tumor-intrinsic factors alone cannot explain the currently identified difference in progression rates seen in the clinic, emphasizing the role of extrinsic factors. The assumption of a central role for the microenvironment (ME) is also supported by the fact that the BM ME is known to affect the differentiation, proliferation, and survival of aberrant PCs. Recent studies have shown that the MM BM ME exhibits quantitative and qualitative alterations in immune, mesenchymal, and dendritic cells. Dysfunction of T-cell subsets and natural killer (NK) cells within the PB and the BM as well as the suppression of myeloid-derived suppressor cells and tumor-associated macrophages in the BM ME have been reported in several studies.13, 14, 15 More importantly, some of these cellular changes, including dysregulation of immune and mesenchymal cells, have been linked to relapse and treatment resistance of MM.16, 17, 18 Previous studies addressing the role of the BM ME using single-cell RNA sequencing (scan-seq) to compare the BM of healthy subjects to those with PC dyscrasias demonstrated that changes in the immune system occur early at the MGUS stage, where an increase in CD16+ monocytes, NK cells, T cells, and precursors is seen. However, the authors could not identify consistent changes across the different clinical stages of the disease, probably reflecting the high degree of variation in cell proportions at all stages. To account for the high degree of variability in cellular populations and improve the detection of changes between early and late clinical stages of PC dyscrasias, we performed scorns-seq using BM samples from patients with MGUS, SMM, or newly diagnosed MM (NDMM) and analyzed the data in conjunction with other available data sets. Furthermore, we studied T-cell clonality at the single-cell level for the first time. We showed that the transition of the precursor stages from MGUS and SMM to MM is associated with significant alterations in the BM ME, in particular, a decrease of CD4 cells, an increase of tumor-associated CD8+ T cells and T-regulatory (Treg) cells, and an expansion of monocytes and polyclonal memory (Mem) B cells.
Primary BM and PB samples were collected from patients with MGUS (n = 9), SMM (n = 7), and NDMM (n = 10) at the University of Arkansas for Medical Sciences (UAMS), Little Rock, AR, and the University of Heidelberg (UoH), Baden-Württemberg, Germany. Informed consent was obtained from all patients in accordance with the declaration of Helsinki, and the study was approved by the institutional review boards of UAMS and UoH. CD138-depleted BM samples were viably frozen in dimethyl sulfoxide at a final concentration of 10% and processed for scRNA-seq and T-cell receptor (TCR) sequencing as described below.
scRNA-seq using the 10× Genomics Single Cell 5′ version 1 (UAMS) and version 1.1 (UoH) kit was performed on the PC-depleted mononuclear fraction of BM aspirates from patients with MGUS (n = 8), SMM (n = 7), and NDMM (n = 10) (supplemental Table 1). Cryopreserved samples were thawed at 37°C and washed twice with ice-cold 1× phosphate-buffered saline. Single-cell capture (target, 3000 cells), reverse transcription, library preparation (expression and TCR), and paired-end sequencing were performed according to the manufacturer’s protocol. All BM samples were further investigated by 8 color flow cytometry using CD138, CD38, CD45, CD19, CD56, CD20, CD27, and CD81 to distinguish B-, T-, NK-, and immature B-cell subsets as well as monocytes. In addition, matched PB samples were used to determine the degree of correlation between B-, NK-, and T-cell BM populations in the BM and PB.
Preprocessing of the 10× scRNA-seq and TCR data was performed with CellRanger 3.1.0 (UAMS) and 5.0.0 (UoH) using standard parameters and hg38 as a reference. Count matrices were loaded into R (version 4.0.0) using standard Seurat parameters and were annotated for patient and disease stages. Cells with >10% mitochondrial RNA or <400 or >3000 detected genes were removed. Ultrahighly expressed genes such as immunoglobulin genes were also removed. Cell doublets were predicted using Scrublet, and cells with a prediction score >0.25 were filtered out. The UAMS and UoH data sets were integrated using Harmony (https://www.nature.com/articles/s41592-019-0619-0). Cell type assignment was performed with the multimodal reference mapping approach from Seurat. For this approach, the CITE-seq BM data set from Stoeckius et al was used as reference. Because the reference data set was log normalized, this normalization approach was also applied to our data set. In brief, the cell type assignment of each sample was as follows: anchors were defined between the reference and each query sample, and then each sample was individually mapped to the reference. In the next step, all annotated samples were merged, as they were integrated into a common reference space and then visualized. The cell type annotation was verified by established marker genes.21, 22, 23, 24, 25, 26 To focus on the tumor ME, PCs and erythroid progenitor cell clusters were removed from the data set. In addition to the cell type assignment, the marker genes SDC1 (CD138) and HBA1 were used for PCs and erythroid progenitors, respectively. The remaining data set consisted of 62 044 cells and varied between 300 and 4000 cells per sample (median, ∼1821 cells). For the specific T-cell analysis, cells classified as T cells were extracted from the data set and clustered separately using k-nearest neighbor clustering with a resolution of ×0.3. TCR data were analyzed with the R-package scRepertoire (https://f1000research.com/articles/9-47/v1), and these additional data were added as metadata to the T-cell Seurat object. Differential expression and gene expression signature analyses were performed using the standard Seurat functions Find(All)Markers and AddModule Score (https://satijalab.org/seurat/index.html). Signatures were calculated for all T cells to determine the level of exhaustion (TIGIT, HAVCR2, CTLA4, PDCD1, LAG3, and LAYN) and cytotoxicity (NKG7, CCL4, CST7, PRF1, GZMA, GZMB, IFNG, and CCL3).
Preprocessed scRNA-seq data from Zavidij et al were downloaded from Gene Expression Omnibus (GSE124310) and analyzed similarly to our data set. Here, all cells with >10% mitochondrial RNA, <200, and >5000 detected genes or a doublet score >0.3 were removed. All other steps including normalization, cell type assignment, and expression signature analysis were performed as described above. The scRNA-seq and TCR-sequencing data of this study have been deposited in the database of Genotypes and Phenotypes (accession number phs002756v1.p1) (UAMS) and the European Genome-phenome Archive (accession number EGAS00001006090) (UoH).
Using a previously published multimodal reference mapping approach, we investigated a total of 62 044 cells in the UAMS and UoH data set, which were classified into 25 subpopulations based on the expression of known marker genes (Figure 1A). T cells were by far the largest subpopulation with 27 621 of the total 62 044 cells (44.5%) and were clustered into 10 distinct T-cell subpopulations (Figure 1B). Figure 1C shows the distribution of T-cell populations by disease stage and per patient. A striking interpatient heterogeneity within each disease stage is evident, particularly in patients with NDMM. To account for this, we increased the sample size by performing a meta-analysis with samples from a previously published data set (Zavidij et al), which included ∼19 000 cells from 5 patients with MGUS, 11 patients with SMM, and 7 patients with MM. Changes in proportions of cell subsets from MGUS to NDMM are shown in Figure 2A. Other changes are shown in supplemental Figure 1 (MGUS vs SMM, supplemental Figure 1A; SMM vs NDMM, supplemental Figure 1B; MGUS vs advanced stages [SMM/MM], supplemental Figure 1C; and precursor stages [MGUS/SMM] vs NDMM, supplemental Figure 1D). Furthermore, we show the cell type–defining expression per disease stage in Figure 2B and per individual sample in supplemental Figure 2.
T cells were divided into 2 main clusters, consisting of CD4 and CD8 T cells (Figure 1B-C). CD4 cells did not show any significant alterations in the meta-analysis, although CD4 Mem cells, characterized by the high expression of IL7R, PLP2, and Fos seemed to have an overall decrease in NDMM compared with the precursor stages, which was seen in both data sets (P ≤ .05 in the UAMS and UoH data set), although not quite significant in the meta-analysis (P = .12) (Figure 2A). A third CD4+ subset, identified as Treg cells based on the high expression of CD25, FOXP3, and LGALS3, increased in MM compared with MGUS and SMM, which was seen in both data sets and the meta-analysis, albeit the results did not achieve statistical significance, highlighting the extensive variability within this cellular population (Figure 2A). Within CD8 T cells, we observed 6 distinct subpopulations including CD8-naïve cells, CD8 effector 1 (Eff 1) and Eff 2 cells, CD8 Mem 1 and Mem 2 cells, and mucosal-associated invariant T (MAIT) cells (Figure 1B). CD8 Eff 2 cells (GZMH, CCL3, CCL4, and XCL2high) were the only CD8 T-cell subset that showed an increased trend in NDMM proportion compared with MGUS (P = .08 in the meta-analysis) (Figure 2). Intriguingly, the CD8 Eff 2 phenotype has previously been associated with tumor-associated CD8+ T cells, which have been shown to promote tumor proliferation in other cancers., In contrast, changes in CD8 T cells with a cytotoxic phenotype (CD8 Eff 1 cells; GZMH, GZMB, CD53+/CD45RAhigh) were more subtle, and an overall trend toward decreased levels was seen only from MGUS to SMM (supplemental Figure 1A). MAIT cells (CD8/CD4low and KLRB1 [CD161high]) diminished significantly from the precursor stages (MGUS/SMM) to NDMM (P < .5 in the meta-analysis) (supplemental Figure 1D). This decrease was specifically seen from the SMM to the MM stage, which was evident in both data sets and the meta-analysis (P = .07) (supplemental Figure 1C). This is of interest because MAIT cells have recently attracted attention owing to their cytotoxic function and potential for immunotherapeutic targets.,
B cells were divided into naïve B cells and Mem B cells. Mem B cells showed a striking expansion from MGUS to NDMM in both data sets and the meta-analysis (P ≤ .05) (Figure 2). The expansion was most evident and significant from the SMM to MM stage in the meta-analysis (P ≤ .01) (supplemental Figure 1B). This is of interest because previous reports have suggested that Mem B-cell populations could contain an MM stem/progenitor population; however, we did not detect the clonal immunoglobulin heavy chain rearrangement in Mem B cells. Monocytes were distinguished into CD14- or CD16-expressing cells. Zavidij et al previously showed that MM and its precursor stages are enriched in CD16 monocytes compared with healthy BM samples. Our data further propose that enrichment of this cell population was higher at the MM stage than at the precursor stage (P = .09 in the meta-analysis) (supplemental Figure 1D). Intriguingly, we showed that the proportion of CD16 monocytes decreased initially from the MGUS to SMM (supplemental Figure 1A), with a significant increase from SMM to NDMM (P < .05 in the meta-analysis) (supplemental Figure 1B). Interestingly, CD14 monocytes followed a similar trajectory, with an initial decrease in both data sets from MGUS to SMM (P < . 05 in the meta-analysis) (supplemental Figure 1A), with a subsequent increase from SMM to NDMM (P < .05 in the meta-analysis) (supplemental Figure 1C).
To validate the compositional changes seen using scRNA-seq, we performed simultaneous phenotyping using 8-color flow cytometry on BM aspirates collected from 21 patients with MGUS, 18 patients with SMM, and 20 patients with MM. This cohort also included patient samples analyzed using scRNA-seq and used additional patient samples to extend the patient cohorts. Overall, there was a good correlation with a cosine similarity factor of 0.8 or higher in B, T, and NK cells and monocytes (supplemental Table 2). As shown in supplemental Figure 3A-D, we observed an overall decrease of T cells and B cells, whereas NK cells and monocytes increased during the transitioning from MGUS to MM. Again, this was not observed uniformly in all the patients, even not in those with paired precursor and MM samples (n = 8, data not shown), and hence, these changes were subtle and not significant. We further analyzed CD4 and CD8 T cells as well as NK cell distribution in the PB and observed that compositional changes in the BM were not always mirrored in PB. This was particularly true for CD4 and CD8 T cells, where there were no substantial proportional differences between MGUS, SMM, and MM in PB (supplemental Figure 4A-C). There was a trend toward a mild continuous increase in NK cells from MGUS through SMM to MM, similar to the increase in NK cells in the BM (in the UAMS and UoH samples).
In an attempt to identify whether changes in the T-cell activity contributed to progression, we determined the degree of cytotoxicity and exhaustion within each T-cell subset using previously published expression signatures and compared our results with the same external data set (Figure 3A-D). The cytotoxicity signature, characterized by expression of GZMA, GZMB, and NKG7, among other genes, was highly prevalent in the CD8 Eff subgroups (CD8 Eff 1 and Eff 2 cells) as well as CD8 Mem 2 and γδ T cells, whereas it was lowest in Treg, CD4 Mem, CD4 naïve, and CD8 naïve cells (Figure 3A-B). In the UAMS/UoH and external data set, we saw a trend toward increased cytotoxicity in the CD8 Eff subgroups, in particular, the Eff 1 subtype, at the MM stage compared with MGUS and SMM, but none of the results reached statistical significance. We then examined a T-cell exhaustion signature, including PD-1, TIGIT, and TIM-3 (Figure 3C-D). We again observed an overlap of T-cell subtypes with the highest expression of the signature (Treg, CD8 Eff 1 and Eff 2, and Mem 2 cells) and those with the lowest expression (MAIT cells, CD4 Mem and naïve cells, and CD8 naïve cells) in both data sets. We did not observe a distinct increase in the exhaustion signature in any of these T-cell subtypes during MM development. In fact, there was some indication that the expression of exhaustion markers was decreased in MM in some T-cell subsets, particularly Treg cells and CD8 Eff 1 and Eff 2 cells, albeit the results were difficult to interpret, given the many outliers.
T-cell clonality, as a proxy for antigen-driven T-cell expansion, has long been used as a marker of tumor reactivity, and increased levels of clonality in MM have been associated with improved clinical outcome. Yet, the implications of T-cell clonality on MM progression from its precursors remain unknown. Here, we divided clonal cells into hyperexpanded (>5% clonal cells), expanded (1%-5% clonal cells), or nonexpanded (<1% clonal cells) T-cell subsets. Because TCR data were not available for the external data set, we performed a pilot study with our data and showed that the mean proportion of clonal (expanded and hyperexpanded) T cells was similar in MGUS (19.01% ± 18.15%) compared with SMM (14.78% ± 6.7%) and MM (19.5% ± 21.26%) (Figure 3E, left panel). Again, we noted an extensive heterogeneity of clonal distribution between T-cell subsets and patient samples at the same disease stage. The vast majority of hyperexpanded and expanded T cells were identified within the CD8+ compartment, in particular the CD8 Eff 1, Eff 2, and Mem 2 cell subsets, with some clonal expansions also observed in MAIT cells (Figure 3E, right panel). There was a trend toward decreased hyperexpanded and expanded T cells within the CD8 Eff 1 and Mem 2 subsets, albeit not statistically significant. To understand how clonal cells might contribute to disease manifestation, we analyzed the differences in expression between clonal (hyperexpanded and expanded) and nonclonal cells within the CD8 Mem 2 cell subset, which had the largest clonal T-cell populations (Figure 3F). Clonally expanded CD8 Mem 2 cells were enriched for markers of cytotoxicity, including GZMB, GNLY, and KLRD1 (cell cytotoxicity coreceptor CD94). In contrast, nonexpanded CD8 Mem 2 cells showed increased expression of markers associated with inflammation (DUSP2, JUNB, and LTB) and immune-aging (GZMK). Furthermore, CD27, a tumor necrosis factor receptor family member and T-cell costimulatory molecule, enhances TCR-induced T-cell expansion. Therefore, the upregulation of CD27 might indicate that nonclonal CD8 Mem 2 cells try to counteract their loss of expansion.
Early intervention for patients with MGUS or SMM to prevent progression to MM is a promising therapeutic option; however, the mechanisms underlying this progression are still not fully understood. A previous landmark article by Zavidij et al has shown significant alterations in the BM ME of patients with PC dyscrasias compared with healthy subjects; yet, consistent changes across the precursor and clinical stages of the disease remain largely elusive, reflecting the high degree of variation in cell proportions at all stages. In an attempt to investigate the cellular and molecular patterns of the BM ME and immune deregulation that discriminate MM from MGUS and SMM, we combined data sets from UAMS/UoH and the previously published data set by Zavidij et al, resulting in a total of 12 patients with MGUS, 18 patients with SMM, and 18 patients with MM. To our knowledge, this is the largest data set to examine BM ME changes at the scRNA-seq level to date. We further determined the clonal expansion of T-cell subsets through progression to MM from the precursor stages, which, to our knowledge, has not been done before. Our results show that the antitumor immune response tends to decline from MGUS to SMM to MM, as shown by a general decrease in naïve and Mem CD4 T cells and an increase of Treg cells together with CD8 Eff 2 cells. CD4 T cells play a critical role in developing and sustaining effective antitumor immunity and are crucial in orchestrating the immune response through activation and maintenance of cytotoxic CD8 T cells, secretion of Eff cytokines, and direct cytotoxicity against tumor cells. Recent studies in MM have shown that CD4 T cells elicit effective anti-MM responses, and that decreased CD4 T-cell counts are associated with adverse prognosis and diminished treatment responses,,, highlighting an important functional role of CD4 T cells. CD8 Eff 2 cells were the only CD8 T-cell subset that significantly increased (P ≤ .1) in MM compared with its precursor stages. This cell subset is characterized by high expression of GZMH, CCL3, CCL4, and XCL2, an expression pattern associated with tumor infiltrating CD8 T cells that have been shown to express significant levels of key inhibitory receptors and promote tumor proliferation in other cancers., Intriguingly, CD8 Eff 2 cells showed high expression of markers associated with cytotoxicity that did not change significantly from MGUS to MM, suggesting that these cells remain in a functionally conserved effector state. Exhaustion markers were also highly upregulated in CD8 Eff 2 cells, indicating that along with the high expression of cytotoxic markers, some degree of hyporesponsiveness coexists, which needs to be further elucidated, particularly in the light of clinically available checkpoint inhibitors. We identified a substantial upregulation of Treg cells (FOXP3, IL2RA, and CTL4A) in MM compared with MGUS, a phenomenon that has been described previously and is strongly associated with MM progression.,, Furthermore, we observed an expansion of Mem B cells in MM compared with MGUS, which was observed across both data sets. However, this alteration was not accompanied by an increase of clonotypic B cells as previously suggested, and it remains unclear whether the increase of B-cell populations in MM is a reactive event or a crucial contribution to MM development. Furthermore, the increase of CD14+ and CD16+ monocytes in MM compared with that in its precursor stages is of interest. Previous reports have similarly shown an increase of this cell subset during MM progression and suggested a supportive role of monocytes in MM growth. There is further evidence that monocytes stimulate the development of MM bone disease by releasing factors that promote osteoclastogenesis. These results underscore a potential crucial role of monocytes in the progression of MM. Notably, even after combining data sets to increase patient numbers, there is evident interpatient heterogeneity. The resulting discrepant results in some cell populations between the UAMS/UoH and Zavidij et al data set could suggest that some of the immune populations possibly do not play a crucial role in transitioning from MGUS to MM and/or that some alterations are rather prognostic or reflective, as reported in multiple reports, than causative.41, 42, 43, 44 Furthermore, it is possible that the stratification by PC genotype rather than disease status could yield more uniform data. This is of particular interest because the occurrence of clone-specific alterations in the BM ME has been previously reported. However, due to the low PC infiltration in precursor stages (particularly MGUS), genotypic data were not available for all samples, and the analysis could not be performed. The addition of more samples in the future will hopefully reveal more significant trends. In addition, examination of sequential paired samples might overcome the observed interpatient heterogeneity. In a next step, we explored T-cell clonality using scRNA-seq in MM and its precursor stages, which, to our knowledge, has not been investigated previously. T-cell clonality, a marker of antigen-driven clonal T-cell expansion, has been associated with improved outcomes in MM and has been shown to correlate with longevity in healthy adults.,, We showed that there is no uniform change of clonal T-cell proportions between precursor and MM stages, suggesting that clonality is more a prognostic marker, which is not bound to disease stage but possibly rather to tumor burden as previously reported. The vast majority of clonal T cells were encountered within the CD8 subsets and had significantly higher expression of cytotoxic markers compared with their polyclonal counterparts, which were enriched for markers of inflammation and aging. These results suggest that T-cell expansion has an immune regulatory role in MM and that the loss of clonality could be directly linked to a decrease in the immune response. However, the reason for the loss of T-cell clonality in MM is not well understood. Although some reports have pointed to increased senescence and exhaustion as possible explanations,, we did not observe a significant increase of exhaustion markers in most T cells in MM compared with the precursor stages. Taken together, we profiled immune alterations in the progressing disease stages of PC disorders and showed some overlapping and significant changes in immune regulation in the independent data sets. Challenges remain with the interpatient heterogeneity in the BM ME that we identified, which can be hopefully achieved in the future by increasing the number of patients investigated and/or by obtaining serial samples from single patients who progress to MM. Determining the early immune events that lead to MM progression will enable us to stratify patients by the risk of progression and generate therapeutic opportunities for early intervention.
Conflict-of-interest disclosure: The authors declare no competing financial interests. |
||
PMC9647437 | Zhang Tingting,Zhou Xiuli,Wang Kun,Sun Liping,Zhuang Yongliang | A review: extraction, phytochemicals, and biological activities of rambutan (Nephelium lappaceum L) peel extract | 31-10-2022 | Fruit,Drying methods,Purification,Bioactive compounds,Functional food | Rambutan (Nephelium lappaceum L.) peels are produced during application process. Drying methods of rambutan peel, including open sun, oven, oven vacuum, and freeze-drying, have been describes in this study. The extraction technologies of dried rambutan peels were reviewed, such as maceration and hot extraction, microwave-assisted extraction, ultrasonic-assisted extraction, and supercritical fluid extraction. The phytochemicals of rambutan peel extracts were analyzed, and the purification and stability of geraniin was reviewed. Rambutan peel extracts exhibit wide bioactivities in vitro and in vivo, and these bioactivities depend chiefly on the phenolic contents and profiles in the different extracts. The safety of rambutan peel extracts was analyzed. In addition, rambutan peel extracts could be used as important components to make different products, which are potentially applied in food, medicine, and cosmetic. However, the extracts efficiency must be further increased using some emerging technologies. Furthermore, the bioactive mechanism and bioavailability of the extract in human system should be further evaluated. | A review: extraction, phytochemicals, and biological activities of rambutan (Nephelium lappaceum L) peel extract
Rambutan (Nephelium lappaceum L.) peels are produced during application process. Drying methods of rambutan peel, including open sun, oven, oven vacuum, and freeze-drying, have been describes in this study. The extraction technologies of dried rambutan peels were reviewed, such as maceration and hot extraction, microwave-assisted extraction, ultrasonic-assisted extraction, and supercritical fluid extraction. The phytochemicals of rambutan peel extracts were analyzed, and the purification and stability of geraniin was reviewed. Rambutan peel extracts exhibit wide bioactivities in vitro and in vivo, and these bioactivities depend chiefly on the phenolic contents and profiles in the different extracts. The safety of rambutan peel extracts was analyzed. In addition, rambutan peel extracts could be used as important components to make different products, which are potentially applied in food, medicine, and cosmetic. However, the extracts efficiency must be further increased using some emerging technologies. Furthermore, the bioactive mechanism and bioavailability of the extract in human system should be further evaluated.
Rambutan (Nephelium lappaceum L.), which belongs to the family of Sapindaceae, is an important tropical commercial fruit in Southeast Asia, Australia, South America, and African countries. In recent years, rambutan is gaining increasing attention and widely acceptance due to its sweet, juicy and exotic appearance. The color of rambutan ranges from green, red and yellow to orange yellow, and the normal size is approximately 3–4 cm in diameter and 3–6 cm in length (Muhamed et al., 2019). The rambutan fruit is composed of peel, pulp, seed, and embryo, which account for 45.7%, 44.8%, 9.5%, and 6.1% of the total dry weight, respectively (Jahurul et al., 2020). Rambutan is often used for fresh consumption and industrially processed into jams, juices, canned fruit, jellies, marmalades, and spreads. These processing forms produce a large amount of peel and seed by-products (Mahmood et al., 2018; Rakariyatham et al., 2020). Therefore, taking advantage of these by-products in industrial applications is very important. In recent years, many studies have reported on rambutan peels. The chemical composition of rambutan peel is cellulose, hemicellulose, and lignin, in which the cellulose content is 24.28%, the hemicellulose content is 11.62%, and the lignin content is 35.34% (Hernández-Hernández et al., 2019). In addition, some mineral contents in the rambutan peels were reported in previous studies. Meanwhile, many studies revealed that rambutan peel contains a high content of phenolics (Hernández et al., 2019). Phenolics are well known to be important secondary metabolites that determine the sensory and nutritional quality of plant products. Phenolics play various roles in the overall sensory appearance of food, ranging from color to taste and mouthfeel properties, and they might also indirectly impact aroma perception. Moreover, phenolics can contribute to health benefits related to the dietary ingestion of plant products (Macêdo et al., 2021). Previous studies showed that rambutan peel extracts have anti-inflammatory, antioxidant, antimicrobial, antibacterial, anti-osteoporosis (OP), antiphotoaging, antiproliferative, antihyperglycemic, and antidiabetic activities. The bioactivities of rambutan peel extracts depend mainly on their phenolic contents and compositions. Furthermore, many factors can affect the phenolics and their bioactivities of rambutan peels from different sources, including cultivars, climate, harvest, soil, and extractive process. Numerous studies exhibited that rambutan peels from different localities have different phenolic contents and compositions, leading to different bioactivities (Hernández-Hernández et al., 2019; Jahurul et al., 2020). In this review, the drying and extract methods of rambutan peels will be described. In particular, phytochemicals especially ingredient content and type of phenolics in rambutan peels will be illustrated. Finally, the different bioactivities of different rambutan peel extracts will be presented. The overview is shown in Figure 1. This study will provide theoretical basis and technical support for the utilization of rambutan peels.
Previous studies have shown that the selection of drying method and its parameters could influence the chemical and biological activities, because differences in chemical contents and compositions were observed in the same material with different drying methods (Vidinamo et al., 2020). Open sun, oven, oven vacuum, and freeze-drying methods have been used in the drying process of rambutan peel, as shown in Table 1. Open sun drying is the oldest, most popular, cheapest, and free method for the drying of plant, especially in the tropics and subtropics. However, the slow drying rate of open sun drying results in product exposure to environmental pollution, time consumption, and weather dependence. Open sun drying of different plants usually takes few hours, even to few days, and it is affected by the product characteristics and drying conditions, including physical structure, humidity, temperature, and air velocity. Thinkratok et al. (2014) reported that the rambutan peel was dried at room temperature for 2–4 days. Li, Sun, and Zhuang (2021) studied the drying of rambutan peel at open sun condition, which was dried for 5 days. Oven drying is a low-temperature convection or forced air oven, which is mainly used in laboratory environment. An oven is an independent device with a heat source, a fan for circulation, and multiple trays for drying various foods at one time. As shown in Table 1, rambutan peel drying was investigated using an oven drier at different temperatures from 30 °C to 65 °C in previous studies. Li et al. (2021) compared different oven temperatures of 40 °C, 60 °C and 80 °C and found that the dry time of these temperatures were 31, 25, and 20 h, respectively. In this study, the total phenolic contents at 40 °C, 60 °C and 80 °C were 227.87, 241.83 and 185.83 mg/g, respectively, which showed that different dry temperatures had an obvious effect on the yield of phenolics from rambutan peel. Freeze drying is a dehydration process in which a solvent or suspension is crystallized at a low temperature, and then ice crystal sublimates from solid state to vapor state. During the freezing process, under the conditions of low temperature and lack of oxygen, the material is cooled to a frozen state, and the frozen water is removed by sublimation, which makes the dried raw materials maintain high biological, chemical and physical properties, and reduces the amount of bioactive compounds. Different freeze-drying technique and conditions could affect ice formation and morphology of dehydrated matrices (Harnkarnsujarit et al., 2015), leading to the yield of phenolics from the rambutan peel. Mota, Morte, Silva and Chinalia (2020), Phuong et al. (2020c), Li et al. (2021), and Thitilertdecha et al. (2010) used freeze drying to prepare rambutan peel. Phuong et al. (2020c) and Li et al. (2021) compared oven drying and freeze drying of rambutan peels. Phuong et al. (2020c) reported that the yield of phenolic contents using freeze drying was 30 g GAE/100 g dry weight (dw), which was higher than that of oven-dried rambutan peel (17 g GAE/100 g dw). Li et al. found that the total phenolic content of freeze drying was 278.33 mg/g dw, which was significantly higher than those of oven drying. These results showed that freeze drying could obtain a higher retention of bioactive components and bioactivities, which was in accordance with previous conclusions (Harnkarnsujarit and Charoenrein, 2011; Siol et al., 2022).
The extract rate of each bioactive compound from plant is widely known to depend on the type of extract method applied (Lucia et al., 2020). Therefore, the extraction process of phenolic compounds from plant is the most crucial step in any related research (Garcia-Salas et al., 2010; Gallego et al., 2019). The common factors mainly influencing the extraction processes are substrate properties of the materials, the type of solvent used, extract temperature and time, liquid–solid ratio, and sample particle size (Minh et al., 2020). The type of extract solvent is one of the most important factors affecting the extract efficiency of bioactive compounds. Solvent polarity plays a significant role in improving the solubility of the phenolics (Garcia-Salas et al., 2010). Solvents with different concentrations and polarities have been applied to extract phenolics from rambutan peels. As shown in Table 1, the extract solvents of rambutan peel mainly include water, ethanol, and methanol with different concentrations. In addition, the usage of hydroethanol, hydrochloric acid, and sodium hydroxide as extract solvent was reported, in which hydroethanol allowed for the extraction of both polar and semipolar compounds. Phuong et al., 2020a, Phuong et al., 2020b, Phuong et al., 2020c revealed that the total phenolic content extracted from rambutan peel was 310 mg gallic acid equivalents (GAE)/g when using 80% of methanol. Subramaniam, Radhakrishnan, Chakravarthi, Palanisamy and Haleagrahara (2015) obtained a 41.1% yield of ethanolic (1:10, w/v) extract from rambutan peel. Palanisamy, Ming, Masilamani, Subramaniam, Teng and Radhakrishnan (2008) compared the phenolic yields with extracting solution of water and ethanol (1:10, w/v) concentration and discovered that the extract with ethanol displayed a higher yield. Phuong et al., 2020a, Phuong et al., 2020b, Phuong et al., 2020c used 80% methanol, 50% ethanol, and water as extract solvents to obtain rambutan peel extracts, and the yields of total phenolics were 17.11, 12.25, and 9.17 g GAE/100 g, respectively. Monrroy, Araúz and García (2020) researched the effect of 10 g/L hydrochloric acid, 10 g/L sodium hydroxide, water, 96% ethanol, and 600 g/L hydroethanol as extract solvents on the yield of phenolics from rambutan peel, and the yields were 189, 258, 235, 233, and 315 mg GAE/g, respectively. Gusman and Tsai (2015) studied the effect of ethanol with different concentrations (0%, 40%, 60%, 80%, and 95%) on the yield of total phenolics of rambutan peel. The results showed ethanol concentration significantly affected the yield of total phenolics, and 40% of ethanol markedly demonstrated the highest extraction efficiency. Chaiwarit et al. (2021) compared the effect of water, 95% ethanol, and 60% ethanol on the phenolic extract yield of rambutan peel, and the yields were 26.68%, 26.92%, and 34.92%, respectively. Sun et al. (2012) utilized three different solvents, namely, deionized water, 60% ethanol, and 60% methanol, to extract the phenolics of rambutan peel, and the result showed the ethanol extract had significantly higher soluble phenolic contents than the water and methanol extracts. The above literature reports showed that the differences in phenolic yield could be explained by the difference in solvent polarity. The polarity of each type of phenolics is determined by the difference in carbon skeleton and the type and quantity of substituents. The different solvents have different polarities, and the polarities of common solvents were water > methanol > ethanol > acetone. The different solvents can selectively extract different hydrophobic or hydrophilic phenolics from rambutan peel, thus highlighting the importance of studying and determining the optimal extract solvent for each sample type. However, rambutan peels from different sources in the process of phenolic extract have different requirements for solvent polarity, which may be related to the forms, compositions and contents of phenolics in the different rambutan peel. Except for the difference in extract solvent, the extract temperature, time, and liquid-to-solid ratio also differed in the extraction process of rambutan peel. As shown in Table 1, the extract temperature ranged from room temperature to boiling, the time ranged from 3 min to 36 h, and the liquid-to-solid ratio ranged from 2:1 to 30:1 according to the previous extract data of rambutan peel. These extraction processes need to be optimized to increase the yield of extract. Response surface method (RSM) is a statistical tool for experimental modelling, which is often used to optimize the extraction process of bioactive compound. RSM can reduce the experimental runs, optimize the experimental processes, and check the interaction between the extract factors without reducing its quality (Box et al. 1978). Therefore, RSM could give the best value of the process factors to achieve maximum extraction or good response. Boyano-Orozco et al. (2020), Maran et al. (2017), and Sun et al. (2012) optimized the extract condition of rambutan peel by using the RSM design. Previous studies showed that the extract methods of rambutan peel included conventional and new extraction techniques. The conventional extraction method is maceration and hot extraction, including reflux, shaking, blending, and stirring. Previous studies mostly chose conventional extraction method to prepare rambutan peel extraction. However, maceration and hot extraction usually takes a long time and consumes many solvents, resulting in low efficiency/yield and quality. In recent years, the demand for new extract technologies, including microwave-assisted extraction (MAE), ultrasonic-assisted extraction (UAE), and supercritical fluid extraction (SFE), have been used to obtain the extract of rambutan peels. The potential implements of these techniques improve extract yield, especially the extract of phenolics from rambutan peel. MAE is identified to be a potential alternative to conventional extract method. MAE is a process that uses microwave energy to heat solvent in contact with the materials and isolate the bioactive compounds from the materials into the solvent. The microwave energy is transferred by molecular interactions through the mechanisms of dipolar rotation and ion conduction. Sun et al. (2012) extracted the soluble phenolic compounds in rambutan peel by using MAE with ethanol as solvent, and the operating parameters of MAE were optimized by RSM. The optimum extract conditions were 80.85% ethanol concentration, 58.39 s extract time, and 24.51:1 liquid-to-solid ratio. Under these conditions, the soluble phenolic yield was 213.76 mg GAE/g DW. Chaiwarit et al. (2021) researched the influence of MAE and maceration method with water, 95% ethanol, and 60% ethanol in obtaining bioactive compounds from rambutan peel. The MAE process operated at a frequency of 2450 MHz at 200 W showed higher efficiency than maceration; shortened the extract time from 18 h to only 3 min; and increased the yields in water and 95% ethanol by 5.13% and 2.08%, respectively. However, the yield in 60% ethanol between MAE and maceration had no significant difference. UAE is a potential extract method that has the advantage of being economical and efficient, with shorter extract time. UAE uses ultrasonic bath with different solvent concentrations, liquid-to-solid ratios, times, temperatures, and ultrasonic frequencies. These factors may vary and need to be optimized in accordance with the type of materials. The phenolics obtained from rambutan peel by UAE is an emerging interest. The optimal conditions, including the type of solvents and their concentration, time, and the temperature for extracting phenolics from rambutan peel by using UAE method were explored by Phuong et al., 2020a, Phuong et al., 2020b, Phuong et al., 2020c. Methanol solution (80%) was found to be beneficial to improve the extraction yields of phenolics due to its polarity and the minimization of the amount of solvent. UAE significantly increased the phenolic yield of rambutan peel by 15% compared with methanolic extraction. Finally, an UAE extraction procedure obtained 55–64 g of frozen powder from 1 kg of fresh rambutan peel, and the total phenolic content in this powder was 250–300 mg GAE/g. Monrroy et al. (2020) studied the extracts of rambutan peel by using different solvents via UAE. The extracts of 1 g of dry rambutan peel were obtained by UAE or boiling for 10 min, with solvents varying from 10 g/L sodium hydroxide NaOH, 10 g/L hydrochloric acid, 96% ethanol, and 600 g/L hydroethanol to aqueous solutions. The phenolic contents in the extracts obtained by UAE ranged from 208 mg/g to 340 mg GAE/g, while the extracts obtained by boiling ranged from 189 mg GAE/g to 315 mg GAE/g. The phenolic yields obtained by UAE with sodium hydroxide, hydrochloric acid, hydroethanol, and aqueous solution were significantly higher than those obtained by boiling. However, the phenolic content of ethanolic extract in UAE decreased compared with that in boiling. Méndez-Flores et al.(2018) obtained rambutan peel extract by UAE, and the best extract conditions were mass/volume ratio of 1:7, extraction time of 10 min, and 10% ethanol/water. In this condition, the total phenolic content was 487.67 mg GAE/g. Maran et al. (2017) obtained bioactive compounds from rambutan peel by UAE accompanied with RSM. The optimal conditions of all the process factors were ultrasound power of 20 W, extract temperature of 50 °C, solid-to-liquid ratio of 1:18.6 g/mL, and extract time of 20 min. Under these conditions, the phenolic yield was 552.64 mg GAE/g. Moreover, Gusman et al. (2015) investigated the optimal condition of extract from two rambutan peels (red and yellow) in Taiwan. This study indicated that liquid-to-solid ratio, solvent concentration, duration, and the method of extraction significantly influenced the recovery of phenolic extracts. The optimal liquid-to-solid ratio was 15:1 when using 40% ethanol, and the extract time was 12 h for conventional extract but only 2 min for UAE. In addition, UAE for 2 min resulted in a higher phenolics recovery than with conventional extract for 12 h, suggesting that UAE could be more effective than conventional method for extracting phenolics from rambutan peel. In addition, it was found that the combination of MAE and UAE technologies showed high potential to extract phenolics. Estrada-Gil et al. (2022) researched the Mexican rambutan peel and found that the yield of MAE/UAE hybrid extraction was an average of 156.96 mg of GAE/g of dry rambutan peel for the hydrolyzable phenolics content, followed by UAE and MAE with 21.32 and 9.48 mg of GAE/g of dry rambutan peel, respectively. Compared with previous studies, the yield of phenolics in rambutan peel was low. However, the study indicated that the MAE-UAE combined technology resulted in a high extraction yield of phenolics, compared to the yields obtained by a single extraction technology. SFE has become an effective method to separate bioactive compounds. The extraction of phenolics by SFE refers to the process of using supercritical fluid to extract bioactive compounds from solid or liquid substances under the condition of higher than the critical pressure and temperature of the fluid. When the fluid pressure is above its critical pressure, the density of the fluid increases and the solubility of the bioactive compounds increases rapidly. When the fluid pressure falls below the critical pressure, the solubility of the bioactive compounds in the fluid decreases rapidly and then precipitates out of the fluid. The separation of phenolics by SFE has been widely studied throughout the last two decades (Katarzyna et al., 2018). Palanisamy et al. (2008) obtained rambutan peel phenolics by SFE. The rambutan peel powder was placed in an extraction vessel, and CO2 was passed through the vessel at a pressure of 300 bar and a temperature of 50 °C for 2 h. The gas flow rate was set at 30 g/min. The soluble fraction was collected in liquid CO2 precipitated at the end of the run. The product obtained was a completely pure extract without liquid organic solvents. Ethanol was used to flush the lines. The obtained yield of the ethanolic extract of rambutan peel by SFE was 17.8%, and the extract had a high phenolic purity of 762 mg GAE/g extract. In summary, the extraction yields of different literatures are very different, which may be caused by different cultivated varieties of rambutan, determination methods of phenolics contents and calculation methods of phenolics yields. Generally, compared with traditional extract technologies, the new extract technologies significantly improved the extraction of phenolics from rambutan peel. Meanwhile, the new extract technologies could realize automation, shorten the extraction time, and reduce organic solvent consumption. MAE can effectively release phenolics because of the heat irradiation, which is produced through the vibration of the water molecules in the medium, but high temperatures can result in a degradation of phenolics when exposed to long time. The non-heating characteristic of UAE makes it more convenient to apply than MAE. According to the previous studies, the UAE extraction technology takes the shortest time and has the highest efficiency of extracting phenolics. However, considering the ultrasonic action area and attenuation, the diameter of extraction equipment should not be too large, which may affect the processing capacity of rambutan peel. In a word, different extraction technologies with optimal condition can achieve the best extraction results. In the process of optimizing extraction, the advantages and disadvantages of different extraction technologies should be fully considered.
Phytochemicals, especially phenolics, in plants are thought to be the major bioactive compounds, having potential bioactivities on health. Phytochemicals are classified in accordance with their chemical structure, from simple structures, such as phenolic acids, to highly polymerized structures, such as tannins. These phytochemicals from plants are currently divided into hydroxycinnamic and hydroxybenzoic acid derivatives, flavones, flavonols, flavanols, and anthocyanins (Tsong et al., 2021). Many previous studies reported the phytochemicals of rambutan peel extracts from different regions in the world. According to the investigation of Thitilertdecha et al. (2010) the three main fractions of phenolics extracted from rambutan peel were identified as ellagic acid, corilagin, and geraniin. In addition, 53.5 mg of ellagic acid, 71.9 mg of corilagin, and 568.0 mg of geraniin were gained from 1 g of methanolic extract, and the sum of these three fractions was 693.4 mg/g extract. Nguyen et al. (2019) analyzed two fractions of rambutan peel extract, including soluble and bound fractions. The phenolics of the two fractions were identified using UPLC-QTOF-MS/MS. The main phenolics were ellagic acid, geraniin, and galloylshikimic acid in the soluble fraction, while ellagic acid, gallic acid, and quercetin hexoside were found in the bound fraction. In addition, 13 compounds extracted from rambutan peel through HPLC/ESI/MS method were found by Hernandez et al. (2017), including apigenin, apigenin arabinoside glucoside, bre-vifolin carboxylic acid, castalagin/vescalagin, corilagin, geraniin, p-coumaroyl glucose, vanillic acid, ellagic acid, ellagic acid pentoside, galloyl-bis-HHDP-hexoside, hexoside, pelar-gonidin, and vitisin A. The main compounds were identified in accordance with peak area, including ellagic acid, corilagin, and geraniin. Besides the ingredients mentioned above, six other ingredients of phenolics from rambutan peel were explored, including gallic acid, isorhamnetin 3-O-glucoside-7-O-rhamnoside, gallic acid 3-O-gallate, galloyl-HHDP-hexoside, pedunculagin, and theaflavin 3,3′-O-digallate. The ingredient content and type of phenolics had otherness when extracted with different solvents through different methods. A study by Phuong et al. (2020c) showed that the main phenolics in methanolic extract were geraniin, ellagic acid, quercetin, rutin, and corilagin. Among them, geraniin had the highest content with two isomers (397 mg/g), followed by ellagic acid and quercetin at 177 and 167 mg/g, respectively. The main phenolic types in the water extract were same as those in the methanolic extract, but the content was less, in which quercetin demonstrated the highest content (186 mg/g), followed by ellagic acid and geraniin (155 mg/g and 137 mg/g, respectively). This study showed that the effect of solvents on the phenolics of rambutan peel extract was obvious. Asghar et al. (2021) prepared rambutan peel extracts by using different solvents in the order of their increasing polarity viz as follows: chloroform < ethyl acetate < acetone < ethanol < methanol < water. Chemical profiles were analyzed by HPLC and LC-MS. Only three compounds were identified by HPLC in the ethyl acetate extract, namely, malic acid, vitamin C, and chlorogenic acid. Three more were found in the acetone extract, including epigallocatechin gallate, catechin hydrate, and quercetin. Furthermore, 54 and 44 compounds were revealed in ethyl acetate and axetone extracts by LC-MS analysis. Thitilertdecha and Rakariyatham (2011) reported the accumulation of phenolics at different growth stages of two rambutan cultivar (Rongrien and Seeechompoo) peels. During fruit maturation, the accumulation of phenolics in the rambutan peel of Rongrien and Seechompoo cultivars continuously improved, until reaching a maximum of 1653 and 733 mg/fruit when rambutans were harvested at 112 and 98 days after full bloom, respectively. Geraniin, corilagin, and ellagic acid were found in the peels of both cultivars. They were quantified, and the major component was found to be geraniin. The accumulation of geraniin, corilagin, and ellagic acid in the peels increased and reached the maximum at the harvest stage. In particular, the contents of geraniin could reach 1011 and 444 mg/fruit for Rongrien and Seechompoo cultivars, respectively. In the previous studies, the phenolics of rambutan peel were analyzed (Sun et al., 2012). Three fractions, including free, soluble conjugate, and insoluble-bound phenolics, were obtained by alkaline hydrolysis, and their contents were 185.12, 27.98, and 9.37 mg GAE/g dry weight, respectively. The soluble extract was obtained by ethanol solvent, and 51 compounds were identified in the extract by using UPLC-Q-Orbitrap-MS2. This extract was purified by NKA-9 resin adsorption technology, and the purification processes increased the total phenolic purity from 579.72 mg GAE/g extract to 877.11 mg GAE/g extract (Zhuang et al., 2017b). The purification process also removed citric acid, quinic acid, ferullic acid hexoside, apigenin glucoside, and kaempferol hexoside. Thirty-nine compounds were identified from the purified extract, including one simple phenolic acid, one flavone, five hydrolyzable tannins, five hydroxybenzoic acids, six ellagic acids and conjugates, 10 flavonols, and 11 flavonols. Geraniin was semi-quantified by gallic acid via UPLC-Q-Orbitrap-MS2 and showed 122.18 mg/g extract, which was the highest among all identified phenolics. Corilagin was quantified using its standard being 7.56 mg/g dry weight. Furthermore, UPLC-QQQ-MS was applied to accurately quantify the contents of geraniin and corilagin by their respective standards (Li et al., 2018). The contents of corilagin and geraniin were 7.87 and 140.02 mg/g, respectively. The contents of corilagin quantified using UPLC-Q-Orbitrap-MS2 and UPLC-QQQ-MS by its standard had no significant difference. However, the content of geraniin was higher than that of UPLC-QQQ-MS quantified by the standard, compared to the semi-quantitative results of UPLC-Q-Orbitrap-MS2 by gallic acid. According to above reports, the phytochemicals of different rambutan peel extracts are mainly ellagitannins, including geraniin, corilagin, and ellagic acid. However, the phenolic profiles of rambutan peels are diverse due to the differences in cultivars, cultivable soil, and methods for extract and analysis.
Geraniin, as a kind of ellagitannin, has attracted much attention (Cheng et al., 2017; Perera et al., 2015). Geraniin was found to be the major constituent of rambutan peel extracts. Thus, rambutan peel could be used as a potential source of geraniin. Large-scale purification of geraniin from rambutan peel extracts should be completed to obtain geraniin with relatively high purity to meet the requirement of geraniin quantity for industrial application. Zhuang et al. (2017b) purified crude rambutan peel extract using different resins. NKA-9 resin was chosen to dynamically purify the rambutan peel extract due to good adsorption ability and desorption ratio. The content of geraniin was from 88.32 mg/g extract to 122.18 mg/g extract. Palanisamy et al. (2011) focused on obtaining geraniin from the ethanolic extract of rambutan peen. Five g of ethanolic extract, obtained from 16.7 g rambutan peel, was first separated on a RP-18 glass column to yield 3 g of yellowish fraction. One g of this fraction was used in subsequent purification on a preparative HPLC. Geraniin was further obtained with 20% acetonitrile. Finally, geraniin with high purity was obtained. Another purification method was reported by Perera et al. (2012). A total of 362 g of crude ethanolic extract was obtained from a weight of 1506 g of dried rambutan peel, and the yield was 24.06%. The crude extract was subjected to reverse-phase C18 chromatography in 20 g batches to isolate geraniin, and geraniin was eluted with acetonitrile:water (10:90) solvent system. The water fraction containing geraniin was crystallized, and the geraniin purified by crystallization was 2.23 g. The total yield of geraniin from the crude extract was 11.15%, and the yield of geraniin from rambutan peel was 2.68%. HPLC analysis of purified geraniin showed that geraniin had a high purity of 97.80%. A minor impurity identified as corilagin constituted 2.20%. The previous study reported a rapid separation and purification method for geraniin from rambutan peel extract (Li et al., 2021). The content of geraniin in freeze-dried rambutan peel was 12.67% (w/w). The extract was purified by medium-pressure liquid chromatography and preparative HPLC. Finally, geraniin with a purity of 95.63% and a yield of 6.00% was obtained. The purity of geraniin was slight lower than that in the study of Perera et al. (2012), but the yield of geraniin had an obvious improvement.
Geraniin, which is rich in rambutan peel extract, is unstable and could easily produce a series of homologous derivatives under the effect of various factors, including alkali, temperature, light, enzymes, microorganisms, and gut bacteria (Espín et al., 2007). The changes in geraniin in previous studies are shown in Figure 2. The stability of geraniin was significantly affected by thermal treatment. Li et al. (2021) studied that the mass concentration of geraniin reduced from 304.48 μg/mL to 252.31 μg/mL after treatment at 60 °C for 10 h, and the percentage of degradation loss was 17.13%. During the first 2 h of heating at 100 °C, geraniin was rapidly degraded, and the percentage of degradation loss was 73.97%. After heating at 100 °C for 10 h, no geraniin was detected. The thermal degradation products of geraniin belonged to phenolics, and seven compounds were mainly identified using UPLC-Q-Orbitrap-MS2, including gallic acid, corilagin, hexahydroxybiphthalic acid, brevifolin carboxylic acid, galloyl-bis-HHDP-glucose, ellagic acid, and brevifolin. Several studies showed that ellagitannins are not intactly absorbed; however, they could be metabolized by the intestinal flora. As an ellagitannin, the initial metabolism of geraniin begins in the stomach, where geraniin is hydrolyzed to free ellagitannins, and a small percentage of geraniins were further metabolized by gut bacteria into smaller metabolites, such as tetrahydroxy-urolithin D, trihydroxy-urolithin C, dihydroxy-urolithin A, and monohydroxy-urolithin B, which are absorbed into the circulation due to their increased lipophilicity (Perera et al., 2015).
In this section, some studies on the bioactivities of different rambutan peel extracts, including antioxidant, antimicrobial, antihyperglycemic, antidiabetic, inhibitory skin aging, anti-OP, anti-inflammatory, and antiproliferative activities, in various tests in vitro and in vivo were reviewed, as shown in Table 2.
Antioxidant activity is widely known to scavenge free radicals, maintain redox balance and metal chelating activity, inhibit enzymatic and non-enzymatic activities, and then regulate oxidant stress. Due to the existence of multiple oxidation mechanisms, no single determination method could accurately reflect the total antioxidant capacity of rambutan peel extracts. Therefore, several antioxidant activity indicators were simultaneously measured after a rambutan peel extract was obtained in previous reports. In addition, rambutan peel extracts could be applied as food preservative due to their antioxidant activities.
Palanisamy et al. (2008) illustrated that the rambutan peel ethanolic extract has a good scavenging free radical activity, comparable to that of vitamin C and much higher than that of grape seed. This study was the first to demonstrate that the ethanolic extract from rambutan peel had high phenolic content, low prooxidant capacity, and good antioxidant activity. The phenolic content of this extract was 762 mg GAE/g, and the IC50 values of Galvinoxyl and ABTs-scavenging activities were 1.7 and 1.7 μg/mL, respectively. In a study by Monrroy et al. (2020), the rambutan peel extract exhibited high antioxidant activity with high inhibitory ABTS∙+ radical cations and DPPH radical activities. The antioxidant activity of rambutan peel extract is higher than those reported about apples, grapes, kiwis, plums, broccoli, garlic, peppers, and spinach. Nguyen et al. (2019) reported the antioxidant activities of two rambutan peel extracts (soluble and bound phenolic extracts), including inhibitory DPPH and ABTS∙+ radicals and FRAP values. The scavenging DPPH activities of the soluble and bound phenolic extracts were 46.38 and 30.87 g TE/100 g, respectively, while their inhibitory ABTS∙+ activities were 54.09 and 42.95 g TE/100 g. In addition, FRAP values at 66.05 and 27.63 g Fe2+/100 g were found in the soluble and bound phenolic extracts, respectively. More antioxidant activity indicators were established by Sun et al. (2012), who prepared free, soluble conjugate, and insoluble-bound fractions by alkaline hydrolysis, and the antioxidant activities of the three fractions were evaluated in five model systems in vitro, including DPPH-scavenging activity, reducing power, lipid peroxidation inhibition activity, OH• scavenging activity, and nitrite-scavenging activity. The results showed that the free fraction had higher antioxidant activities than the soluble conjugate and insoluble-bound fractions. The IC50 values of the free fraction were 4.21, 3.55, 85.53, 19.12, and 17.06 μg/mL in five model systems. Li et al. (2018) reported that rambutan peel extract had good Fe2+ and Cu2+ chelating capacities, with EC50 values of 0.80 and 0.13 mg/mL, respectively. Meanwhile, the extract obviously inhibited the production of hydroxyl radical, with an IC50 of 62.4 μg/mL. Impressively, the effect of this extract on AAPH-induced DNA damage was explored in this study, and the extract effectively inhibited radical-induced plasmid DNA strand breakage. In addition, this extract showed a protective effect on H2O2-induced oxidative damage in HepG2 cells (Zhuang et al., 2017a). It extract could reduce the intracellular level of ROS, increase SOD activity in HepG2 cells, improve oxidative stress defense, and then inhibit cell apoptosis. Thitilertdecha et al. (2010) evaluated the antioxidant activities of rambutan peel extracts obtained using ether, methanolic, and aqueous solvents, respectively. Several potential antioxidant indicators, including free-radical scavenging activity, reducing power, linoleic peroxidation, and β-carotene bleaching, were evaluated. The methanolic fraction had the highest antioxidant activity, as evidenced by the IC50 of DPPH inhibition being 4.94 μg/mL, higher than that in ascorbic acid and BHT. Sekar et al. (2014) reported the inhibitory effects of different rambutan peel extracts on DPPH radical and tyrosinase, and the IC50 of methanolic extract for DPPH and tyrosinase inhibitory activities were 38.88 and 51.44 μg/mL, respectively. According to above studies, different rambutan peel extracts had different in vitro antioxidant activities, which were depend chiefly on their phenolic contents and profiles.
The previous study reported that the rambutan peel ethanolic extract has a protective effect on D-gal-induced aging in mice (Zhuang et al., 2017a, Zhuang et al., 2017b). This extract decreased D-gal-induced oxidative stress in mice by regulating the levels of T-AOC, SOD, GSH-Px, and MDA in serum, liver, and kidney. In addition, this extract significantly reduced histopathological changes in the liver and kidney of D-gal-induced mice. It also had protective effect against lipid peroxidation and accumulation of the liver in obese male Wistar rats, as observed by Setyawati et al. (2015). The research showed that the extract with the dose of 15 and 30 mg/kg bw significantly decreased MDA levels and did not significantly downregulate the expression of PPARγ. Interestingly, rambutan peel extract has the ability to protect the erythrocytes and hemoglobin of rats exposed to cigarette smoke, as reported by Lisdiana et al. (2017). Cigarette smoke is an exogenous free radical, which could damage the structure and function of erythrocyte membrane. The rambutan peel extract could maintain and improve the number of erythrocytes and hemoglobin in the rat blood exposed to cigarette smoke.
Apart from the in-vitro and in-vivo antioxidant activities of rambutan peel extract, the protection effect of the extract on oil by inhibiting lipid oxidative was determined. Phuong et al. (2020b) evaluated the effect of rambutan peel extract on the oxidative stability of soybean oil stored at 4 °C and 30 °C in the dark and light and deep fried with potatoes at 160 °C. The oil mixed with the extract at 1000 μg GAE/g could effectively delay the reaction of oxidation during storage in comparison with the oil without the extract. During frying, the extract could delay the lipid oxidation of oil. The contents of thiobarbituric acid reactive substances of potatoes fried in oil with the extract were much lower than those without the extract. Therefore, rambutan peel extract has good antioxidant effects, and it could effectively inhibit lipid oxidation in oil during storage and deep frying. Furthermore, Mei et al. (2014) reported the effect of rambutan peel extract on the stability of sunflower oil. The crude extract of rambutan peel was purified using silica-packed open column chromatography by increasing the polarity of solvents (ethyl acetate, chloroform, and methanol), and the three subfractions were obtained. One subfraction with a concentration of 300 μg/g was found to have better effect for 2 years of storage period at ambient temperature, comparable with that of tocopherol and BHA. These results indicated that the rambutan peel extract could be a potential source of antioxidants for the stability of sunflower oil.
Rambutan peel extracts have been found to have a wide range of activity against bacteria and microbes. Thitilertdecha et al. (2008) reported the antibacterial activity of different rambutan peel extracts with different solvents (ether, methanol, and aqueous solvent) against eight bacteria, including Escherichia coli (E. coli), Klebsiella pneumoniae (K. pneumonia), Salmonella typhi (S. typhi), Pseudomonas aeruginosa (P. aeruginosa), Vibrio cholerae (V. cholera), Staphylococcus aureus (S. aureus), Enterococcus faecalis (E. faecalis), and Staphylococcus epidermidis (S. epidermidis). The methanolic fraction showed the highest antibacterial activity, and this fraction had antibacterial activity against five pathogenic bacteria, in which S. epidermidis, the most sensitive strain, was inhibited, and the minimal inhibitory concentration (MIC) was 2.0 mg/mL Sekar et al. (2014) indicated that the methonolic extracts of red and yellow rambutan peels had good inhibitory Gram-positive bacterial activity, such as S. pyogenes and S. aureus, and no activity against Gram-negative bacteria, such as E. coli and P. aeruginosa. The highest antibacterial activities of both extracts were found against S. aureus. The methanolic extract of yellow rambutan peels had higher activity than that of red rambutan peels. Tadtong, Athikomkulchai, Worachanon, Chalongpol, Chaichanachaichan and Sareedenchai (2011) showed that the extract has anti-bactericidal activities against S. aureus, MRSA, and S. mutans and no activities against Gram-negative bacteria E. coli and fungus Candida albicans. This finding was similar to the report of Sekar et al. (2014). This extract showed good antimicrobial activity against S. aureus and MRSA, and the diameters of the inhibition zone were >10 mm. In addition, the MIC of the extract against S. aureus and MRSA were 2 and 0.4 mg/mL, respectively. Asghar et al. (2021) studied the antibacterial activity of products extracted from yellow rambutan peel with different solvents against six pathogens, including B. subtilis, P. aeruginosa, MRSA, K. pneumonia, S. pyogenes, and S. enterica. Compared with the extracts obtained using other solvents (chloroform, methanol, ethanol, and water), those obtained using ethyl acetate and acetone showed significant antibacterial activity towards all tested strains. Furthermore, this study reported that the extract computationally inhibited the ATP-binding domain of chaperone, the DnaK of P. aeruginosa and MRSA. In addition, some studies indicated that rambutan peel extract could inhibit the growth of V. parahaemolyticus, V. vulnificus, P. aeruginosa, and C. tetani. These previous studies showed that rambutan peel extracts has good inhibitory Gram-positive bacterial activity and no activity against Gram-negative bacteria. Moreover, the high potency as antibacterial agent against MRSA was identified. Furthermore, Phuong et al. (2020a) studied the antimicrobial activity of methanolic extract from rambutan peel against some Gram-positive and -negative bacteria, and this extract was applied in real food due to its antimicrobial activity. The results showed that the extract had potential inhibitory effect against E. coli, V. parahaemolyticus, V. campbellii, P. aeruginosa, V. anguillarum, S. enteritidis, St. aureus, C. albicans, and L. monocytogenes in vitro. Although food matrices are partially protected against bacteria, the extract inhibited S. enteritidis in raw chicken breast for 14 days at 4 °C in in-situ tests. The extract reduced V. parahaemolyticus by 1.5 log CFU/g in fish during 10 days of cold storage. Moreover, the antimicrobial activity of rambutan peel extract (RPE) can be enhanced by mixing with cinnamon essential oil (CEO), especially when the ratio RPE and CEO was 5:5, the synergist effect against Gram-positive bacterial and Gram-negative bacterial reached the best. The mixture of RPE and CEO is a potential natural antibacterial compound and could be used in food packaging film for prolonging the shelf life of fresh meat and meat products (Khanoonkon et al., 2022).
Diabetes mellitus is a global problem that has a major effect on health, the quality of life, life expectancy, and the healthcare system. One current method to treat diabetes is to inhibit carbohydrate hydrolyzing enzymes, such as α-glucosidase and α-amylase, in the digestive tract to decrease postprandial hyperglycemia by retarding the absorption of glucose. Another method is to inhibit the key enzymes in the polyol pathway, including aldose reductase, which could effectively inhibit the formation of advanced glycosylation end products (AGEs). Rambutan peel extracts could inhibit the activities of several digestive enzymes and decrease the formation of AGEs. First, the antidiabetic activities of rambutan peel extracts were evaluated in vitro. U.D. Palanisamy et al. (2011) showed that the rambutan peel ethanolic extract could inhibit the carbohydrate hydrolyzing enzymes, and the IC50 values of inhibitory α-glucosidase and α-amylase activities were 2.7 and 70.8 μg/mL, respectively. In addition, the extract could inhibit aldol reductase, and the IC50 value was 0.04 μg/mL. The maximum inhibitory activity of the extract for AGE formation was found at the incubation time of 7 days, 43% for rambutan peel ethanolic extract and 38% for green tea. The rambutan peel extract had higher AGE inhibition activity than green tea. Moreover, U. Palanisamy et al. (2011) further evaluated the hypoglycemic activity of geraniin rapidly purified from the extract. The results showed that the purified extract possessed higher hypoglycemic activity, with the IC50 values of inhibitory α-glucosidase and α-amylase activities being 0.92 and 0.93 μg/mL, respectively. Zhuang et al. (2020c) also checked the inhibitory activity of the glycation of crude and purified extracts in vitro. Both exhibited good performance, including the formation of amadori products, dicarbonyl compounds, and AGEs. The purified extract showed stronger bioactivity than the crude extract, possibly due to their differences in phenolic profiles. At a concentration of 10 μg/mL, the inhibitory effects of crude and purified extracts on AGE formation were 32.52% and 44.07%, respectively. Then, the antidiabetic activities were detected in vivo. Ma et al. (2017) evaluated the antidiabetic activity of rambutan peel extract by using a type II diabetic mouse model induced by streptozotocin combined with high-fat diet. The results indicated that the extract regulated the body weight and decreased the fasting blood glucose level in diabetic mice. The extract obviously decreased the serum levels of total cholesterol, creatinine, glycated serum protein, and triglyceride of diabetic mice in a dose dependent manner. The glycogen content in liver was also regulated. In addition, the extract further increased the activity of SOD and GSH-Px in diabetic mice and decreased lipid peroxidation. Histological analysis indicated that the extract could protect the tissue structure of the kidney, liver, and pancreas. Finally, the extract reduced the renal mesangial index and suppressed the expression of TGF-β in diabetic mice. The potential ability of the peel extract in regulating hyperglycemia in diabetes in vitro and in vivo was studied by Subramaniam et al. (2015). The IC50 values of the extract on the inhibitory activities of α-glucosidase and α-amylase were 6.44 and 93.35 μg/mL, respectively, similar to the report published by Palanisamy et al. The in-vivo antidiabetic effects of rambutan peel extract were further evaluated in a high-fat-induced diabetic rat model. The results showed that the rambutan peel extract could reduce the blood glucose level and improve the insulin levels of diabetic rat. Pancreatic histology also showed that rambutan peel extract has healthy pancreas morphology. The activity of the extract was comparable to that of metformin. In addition, Muhtadi et al. (2015) explored the antidiabetic and anti-hypercholesterolemic activities of rambutan peel ethanolic extract, with successive doses of 125, 250, and 500 mg/kg body weight (bw). This study indicated that the extract possessed antidiabetic and antihypercholesterolemia activities at doses of 125–500 mg/kg bw. The blood glucose and cholesterol levels in a dose of 500 mg/kg bw decreased by 61.76% and 60.75%, respectively. The antihypercholesterolemic activity of rambutan peel extract was higher than that of cholestyramine. Lastly, Chung, Ton, Gurtu and Palanisamy (2014) reported that geraniin, which was obtained from rambutan peel, could improve metabolic risks by diet-mimicking metabolic syndrome. This study was the first to show that an orally available geraniin could safely improve many negative pathological sequels of metabolic syndrome. Geraniin at 50 mg/kg bw showed significant therapeutic potential, and it could safely alleviate obesity-induced metabolic dysfunction, including body weights, white adipose tissue depots, organ weights, triaylglycerol, biomarkers of renal and liver dysfunction, insulin resistance, and decreased insulin sensitivity and percentage of beta-cell function. Chen et al. (2020) studied the metabolic effects and possible mechanism of geraniin from rambutan peel in rats with metabolic syndrome induced by high-fat diet. The result showed that geraniin could improve multiple metabolic abnormalities, such as hypertension, impaired glucose and lipid metabolism, and ectopic fat deposition in the visceral fat and liver. Geraniin was found to be comparable to metformin. This finding was similar to the report of Subramaniam et al. (2015). Transcriptomic results showed that geraniin had a profound effect on liver expression. Lipid and steroid metabolic processes were also regulated by geraniin. According to differential transcriptomes, geraniin significantly regulated the expression of mitochondrial genes, which may potentially affect the activity of mitochondria.
Ultraviolet (UV) irradiation is a crucial factor causing skin photoaging. UV irradiation can produce reactive oxygen species (ROS), and excessive UV irradiation can accelerate the production of proinflammation cytokines. ROS and proinflammation cytokines could induce the formation of MMPs, increase MMP contents and activities, and then decrease the formation of collagen in skin. Two methods are often used to protect skin from UV irradiation. One method is to absorb UV irradiation and minimize exposure to radiation; the other is to increase the reduce power in skin. Many reports have shown that rambutan peel extract has two actions for antiphotoaging. Mota et al. (2020) found that rambutan peel extract could be used as a natural additive to enhance the sun protection factor (SPF) of final product. Rambutan peel extract could absorb UVB radiation between the ranges of 290 and 320 nm. Adding 1% concentration of rambutan extract could improve SPF from 0.4 to 11.2 and further increase it to 26.3 when 7.5% of ethylhexyl metoxycinnamate (EHMC) was added. The increase in SPF is due to the synergistic effect between the phenolics of the rambutan peels and the EHMC. In addition, sunscreen formulation containing 1.00% rambutan peel extract showed the potential to minimize the risk of toxicity of synthetic agent and reduce the production cost of sunscreen by 45%. Meanwhile, the utility of rambutan peel extract in skin-aging treatments was reported by Lourith et al. (2017). The elastase inhibitory ability of rambutan peel extract was 31.08% at a tested concentration of 0.25 mg/mL, and the collagenase inhibitory activity was 53.99% at a tested concentration of 0.125 mg/mL. Rambutan peel extract was found to be safe to human skin fibroblasts, and the safe concentration was 0.01 mg/mL. In addition, it inhibited MMP-2 by 23.11% at 0.01 mg/mL. Furthermore, Xiao et al. (2019) studied the protective effects of rambutan peel extract and/or peptide LSGYGP and the additive effect of both on skin photoaging in UV-induced hairless mice. The results indicated that the extract and/or LSGYGP had protective effect on photoaging skin. The extract showed positive effects on the regulation of oxidant stress (antioxidant enzyme activities and glutathione and malondialdehyde contents), inflammatory cytokine levels (IL-1α, TNF-α, and IL-6 levels) and MMP levels (MMP-1, MMP-3, and MMP-9). The histological changes revealed that the extract and LSGYGP had good protective effect on skin tissue and endogenous collagen. Furthermore, the extract and LSGYGP produced an additive effect on skin photoaging in UV-induced mice.
OP is a usual bone disease, which is characterized by low bone density, proneness to bone fragility, and high risk of fracture. According to the report of World Health Organization, with global population aging, about 62% of men and 72% of women over the age of 50 are predicted to suffer from OP or osteopenia by 2022. Zhuang et al. (2020) evaluated the effect of rambutan peel extract on OP by using two models, RANKL-induced RAW264.7 cells and retinoid-acid-induced osteoporotic rats. RAW264.7 cells could be differentiated into osteoclasts by using RANKL. Different concentrations (0.5, 1.0, 2.5, and 5.0 μg/mL) of rambutan peel extract had no significant effect on cell viability. The extract reduced the number of TRAP-positive cells significantly in a dose-dependent manner. In addition, the extract treatment decreased the total TRAP activity in RANKL-stimulated RAW264.7 cells. The extract treatment also significantly improved calcium loss in retinoid-acid-induced osteoporotic rats. The level of serum phosphorus in OP rats was increased, and the levels of total alkaline phosphatase and osteocalcin in serum of OP rats were further decreased. In addition, the extract increased the qualities of the femur and tibia of osteoporotic rats to some extent, such as bone length, bone mineral density, bone maximum load, trabecula relative bone density, and cortical bone area ratio. Histological changes showed that the extract could effectively improve the bone microstructure of OP rats by regulating the trabecular bone separation and cortical bone thickness.
The anti-inflammatory property of rambutan peel extract was tested in an LPS-induced RAW 264.7 cell model. In this model, NO was overproduced and the iNOS level increased obviously. The excessive NO involves a number of events, including the reactions of oxidative stress and inflammation. The production of NO may be derived from the modulation of iNOS. Rambutan peel extract significantly inhibited the NO production and regulated the levels of iNOS mRNA in LPS-induced RAW 264.7 cells, and the activities increased in a dose-dependent manner (Li et al., 2018). Rambutan peel extract also showed positive influence on rheumatoid arthritis, a chronic inflammatory disease that mainly targets the synovial tissue, cartilage, and subchondral bone. The effect of rambutan peel extract on collagen-induced arthritis (CIA) in dark agouti rats was studied by Kumar et al. (2012). CIA rats were given 100 and 200 mg/kg of rambutan peel extract orally from day 25 to day 50. The extract significantly inhibited the physiological, biochemical, and histopathological changes during arthritis rats. In addition, rambutan peel extract could significantly reduce the body weight and paw edema induced by arthritis and reduced the C-reactive protein. After treatment with rambutan extract, the histopathological changes caused by arthritis were significantly regulated. Furthermore, the effect of rambutan extract on the levels of MMP-13 and TIMP-1 was in a dose-dependent manner.
Research revealed that rambutan peel extracts possess antiproliferative activities on human cell lines KB, Caco-2 cells, MDA-MB-231, HeLa human cervical adenocarcinoma, and MG-63 human osteosarcoma cells (Khaizil Emylia et al., 2013; Khonkarn et al., 2010). The yellow cultivar showed slightly better effect than the red cultivar on MDA-MB-231 and MG-63 cells, with IC50 values of 5.42 and 6.97 μg/mL, respectively. However, a higher concentration (≥49.5 μg/mL) of both extracts was needed to decrease the HeLa cell viability. In addition, Ling et al. (2010) reported the anti-proliferative activities of 13 plants native to Malaysia, including rambutan, and the methanolic and aqueous extracts of rambutan peel with doses of 50 and 100 μg/mL had no cytotoxic effects on 4T1 cells and 3T3 cells.
The antihypercholesterolemic activity of rambutan peel extract was evaluated by Suciati et al. (2020). Some main compounds from rambutan peel extract were analyzed as squalene synthase inhibitors by using simulation molecular docking. The docking results indicated that geraniin, corilagin, and ellagic acid could bind with squalene synthase active site and form stable bonds. Geraniin had the lowest binding free energy. In addition, ADMET results revealed that 75% of geraniin, corilagin, and ellagic acid compounds could be absorbed by human digestion. They were well distributed and did not cause liver toxicity. The antihypertensive activity of rambutan peel extract was also reported (Looi et al., 2020). In-vitro studies showed that the geraniin from rambutan peel extract could inhibit ACE. In-silico molecular docking showed that geraniin could form a series of hydrogen and hydrophobic interactions with the active site of ACE and inhibit ACE activity. Furthermore, geraniin could decrease systolic blood pressure in a high-fat diet-induced obese Sprague–Dawley rats. In addition, blood pressure reduction in SHR was reported after oral consumption and intravenous administration of geraniin. The effect of geraniin, extracted from rambutan peel extract, on dengue virus type-2 (DENV-2) was evaluated by Ahmad et al. (2017). The results showed that geraniin inhibited DENV-2 plaque formation, with an IC50 of 1.75 μM. Geraniin decreased viral infectivity and inhibited DENV-2 from attaching to the cells, but it had a slight effect on its penetration. Geraniin was discovered to had high effective at the early stage of DENV-2 infection. According to the molecular docking result, geraniin could interact with DENV E protein at the DIII region, while geraniin was bound to rE-DIII with high affinity.
According to previous studies, phenolics have good bioactivities. Therefore, phenolics could be used as important components to make different products. These products have extensive applications in food, medicine, and cosmetic industry (Khumkomgool et al., 2020). Recently, some reports showed that rambutan peel extract, as an active material, could be made into biofilms, creams, and micro-encapsulations. Go and Song (2020) reported new citrus Junos Pomace cectin (CJP) films combined with different concentrations of rambutan peel extract. These films were able to block out light by reducing light transmittance. In addition, with the dose of the extract increased, the antioxidant activities of the CJP films increased. In addition, the previous study of the authors showed that rambutan peel extract combined with pachyrhizus starch (PS) could be added into tilapia skin collagen (TSC) to improve the properties of TSC films. TSC film containing 10% PS and 0.5% of the extract showed the highest tensile strength. The incorporation of extract and PS improved the thermal stability of TSC films (Zhuang et al., 2019). Yun et al. (2021) found that incorporation of rambutan peel extract (RPE) enhanced the light, water vapor and oxygen barrier abilities, mechanical property, and antioxidant and antimicrobial activities of Chitosan (CS) films. The structural, physical and functional properties of the films were greatly influenced by the addition amount of RPE. CS film with 5% RPE had the highest barrier, mechanical, antioxidant and antimicrobial properties. Moreover, pork wrapped with CS film containing 5% of RPE presented the lowest total volatile basic nitrogen level, thiobarbituric acid reactive substance value, total viable count and the best sensory attributes on the 8th day. The results suggested the potential of CS film containing 5% of RPE as an active packaging material in pork preservation. Chollakup et al. (2020) reported the blending of cassava starch and whey protein isolate films incorporated with cinnamon oil and rambutan peel extract. The effects of native starch, acetylated starch, cinnamon oil, and rambutan peel extract on the film characteristics and antioxidant and antibacterial activities were evaluated. Hydrogen bonding and hydrophobic interaction were observed among protein, starch, and polyphenols. Active compounds reduced water vapor permeability according to the starch types controlling the dispersion of cinnamon oil. Starch promoted the release of phenolics and scavenging radical activities in water and 50% ethanol. In vitro and in real food, the antibacterial activity of the blend films differed, which depended on the release of phenolics and food components, respectively. Rambutan peel extract was used to prepare an antiaging cream by Sekar et al. (2017), due to its antioxidant and antiaging properties. The amount of the extract in the cream was 3% (w/w). The cream was o/w type of emulsion, and the pH was in the range of 4.30–5.20. The physicochemical parameters of this cream, including homogeneity, appearance, odor, spreadability, after feel, type of smear, removal, microbial limit test, and stability, were determined. The results revealed that the formulation of antiaging creams is consistent with the quality of their components, hence useful for consumers. In addition, the cream did not cause redness, edema, inflammation, and irritation in irritancy studies, indicating that it is safe to be used on the skin and has tremendous potential for cosmetic market development. Rambutan peel extract also could serve as microencapsulation. Boyano-orozco et al. (2020) prepared the microencapsulation of rambutan peel extract through spray drying. The best spray drying encapsulating operations were as follows: inlet temperature of 160 °C, outlet temperature of 80 °C, and 10% encapsulating agent dose in the feeding solution. Under these operations, retention and encapsulation levels were higher than 85%, and the values of moisture content, water activity, and Hauser index were 3.95%, 0.25, and 1.42, respectively. The results revealed that the optimized powder of the extract had good solubility and morphological characteristics, and the microcapsule had no ruptures. Microencapsulation of rambutan peel extract could be applied to natural ingredients in food, medicine, and cosmetic industry.
Rambutan peel extract could be used as a natural bioactive agent due to the content, composition, and bioactivities of the phenolics. However, safety is the primary element to be considered before application. Many studies have focused on the safety of different rambutan peel extracts. Thinkratok et al. (2014) studied the safety of rambutan peel extract by a male mouse model. Acute toxicity was evaluated by treating the rats orally with the extract at different concentrations (1000–5000 mg/kg), and the sub-chronic toxicity was evaluated by treating the rats with single doses of the extract (500, 1000, and 2000 mg/kg) daily for 30 days. Food consumption and body weight gain were obviously decreased in acute and sub-chronic toxicity analysis. In an acute toxicity study, the extract did not affect the serum levels of TG, AST, and ALT, and the LD50 value was analyzed to be higher than 5000 mg/kg. In the sub-chronic toxicity study, the extract significantly reduced the plasma TG level and blood urea nitrogen. However, the plasma AST and ALT levels were not altered, and the TC levels did not show any remarkable change. No mortality and toxicity symptoms were found up to 1000 mg/kg/day. However, the mortality rate was 12.5% with a dose of 2000 mg/kg/day. The acute and sub-chronic toxicity of ethanol extract from rambutan peel was evaluated using Sprague–Dawley rats by Subramaniam et al. (2012). In the acute study, a single oral administration of rambutan peel extract with different doses of 50, 200, 1000 and 2000 mg/kg was used in rats for 14 days. In the sub-chronic toxicity study, the extract with two doses of 500 and 2000 mg/kg was administered to rats for 28 days. Neither mortality nor adverse effects were found in rats. No significant differences were observed in the relative organ weights and biochemical analysis. The histology of liver and kidney also showed no obvious changes. Therefore, the lethal dose of the ethanol extract of rambutan peel was higher than 2000 mg/kg. The no-observed-adverse-effect-level of the extract is thought to be up to 2000 mg/kg in rats. The acute and sub-chronic toxicities of rambutan peel extract with higher concentrations than those in the study of Subramaniam et al. (2012) were evaluated using oral administration on Kunming mice and Sprague–Dawley rats, respectively, by Li et al. (2020). In the acute toxicity study, the LD50 of the extract was determined to be higher than 5000 mg/kg bw in vivo. In the subacute toxicity study, the extract showed no obvious adverse effect at doses of 312 and 625 mg/kg bw. However, body weight gain was significantly reduced at a dose of 2500 mg/kg bw of the extract, and the extract at doses of 1250 and 2500 mg/kg bw revealed toxicities to kidney, liver, and spleen in rats based on hematological and biochemical analyses. Furthermore, the extract revealed toxicity on different tissues at 2500 mg/kg bw based on histopathological analyses. A comprehensive acute oral toxicity of geraniin and a geraniin-enriched rambutan peel extract in Sprague–Dawley rats was reported by Moorthy et al. (2019). After a single oral dose, the LD50 values for geraniin and the extract were determined to be 2000 mg/kg bw. The hepatocytes of three rats treated with geraniin exhibited a “foamy appearance.” Therefore, the no-observed-adverse-effect level of geraniin was lower than 2000 mg/kg, while that of rambutan peel extract rich in geraniin was up to 2000 mg/kg. To sum up, the dosage of rambutan peel extract should be carefully selected to improve biological activity and reduce adverse reactions.
Rambutan fruit is widely accepted all over the world, and it is gradually industrialized. The intervention of multi-disciplinary research has promoted the conversion of rambutan peel waste to a healthful ingredient for industries. Rambutan peel has obviously become a good source of food functional components because of its phenolics and bioactivities. However, some challenges in the production and application of phenolics still exist, thereby limiting the rapid expansion of rambutan peel market. The following challenges should be considered in future studies: (1) Although some efficient extraction techniques, such as MAE, UAE, and SFE, were used to prepare the rambutan peel extracts, some emerging technologies, including pulsed electric field, high voltage electrical discharge, instant controlled pressure drop, were not used for the extract of rambutan peels. These technologies may increase the extract efficiency of bioactive compounds by increasing the extract yield, shortening the extraction time, and decreasing energy consumption. Meanwhile, the extract process must be applied for application to a large scale of operation. In addition, producing phenolics with good quality and consistency at the industrial level is difficult, while stability data of phenolics during production and storage procedure are lacking. (2) Producing rambutan peel phenolics with bioactivity and consumer acceptable taste is difficult. The beneficial effects of rambutan peel phenolics in human studies are also lacking. Meta-analysis of scientific evidence related to the health benefits of the phenolics is insufficient. Although many studies have been carried out to determine the bioactivity of phenolics in vitro, the fate of these functional molecules in the gastrointestinal tract and their absorption and bioavailability have not been fully explored. In vivo scientific evidence about the mechanism of action of phenolics and dose–response relationship is insufficient, and data on the bioavailability of various compounds from rambutan peel in the human system are scarce. In addition, scientific data related to the absorption, distribution, metabolism, and excretion of rambutan peel phenolics are not available. If the consumption of phenolics is greater than the recommended amount, scientific data on the risks associated with phenolics should be considered.
All authors listed have significantly contributed to the development and the writing of this article.
Sun Liping was supported by National Natural Science Foundation of Yunnan Province [202101AT070084]. Professor Zhuang Yongliang was supported by 10.13039/100014717National Natural Science Foundation of China [31864021].
No data was used for the research described in the article.
The authors declare no conflict of interest.
No additional information is available for this paper. |
|
PMC9647445 | Sudheer Gupta,Ashvini Yadav,Sam Stubbs,Simon Frost,Kudsia Ansari,Ram Kumar Nema,Shashwati Nema,Debasis Biswas | Genome-wide mutational analysis of Chikungunya strains from 2016 to 2017 outbreak of central India: An attempt to elucidate the immunological basis for outbreak | 04-11-2022 | Chikungunya,Phylogenetics,Outbreak,Genome sequencing,Nanopore | Chikungunya re-emerged in India in 2016–2017, as the first major outbreak since 2006. In our previous study, we undertook partial E1 gene sequencing and phylogenetic/mutational analysis of strains from the 2016–2017 outbreak of Chikungunya in central India and reported important mutations associated with the outbreak. This study was performed to validate the previous findings and to identify key mutations that had emerged throughout the entire genome of Chikungunya virus that could be driving the enormity of this outbreak. The phylogenetic analysis revealed the closeness of our isolates with ECSA genotype, specifically with the Singapore 2015 strain. We found 2 mutations in C and E2 genes, which were present in our isolates but were non-existent during the period of 2010–2016. Furthermore, re-emergence of Arg amino acid in place of stop codon in nsP3 gene and Thr at E2:312 positions was observed after 2011. We also used computational tools to assess the effect of the identified mutations on the T cell and B cell epitopes that could influence the protective immune response against this infection. | Genome-wide mutational analysis of Chikungunya strains from 2016 to 2017 outbreak of central India: An attempt to elucidate the immunological basis for outbreak
Chikungunya re-emerged in India in 2016–2017, as the first major outbreak since 2006. In our previous study, we undertook partial E1 gene sequencing and phylogenetic/mutational analysis of strains from the 2016–2017 outbreak of Chikungunya in central India and reported important mutations associated with the outbreak. This study was performed to validate the previous findings and to identify key mutations that had emerged throughout the entire genome of Chikungunya virus that could be driving the enormity of this outbreak. The phylogenetic analysis revealed the closeness of our isolates with ECSA genotype, specifically with the Singapore 2015 strain. We found 2 mutations in C and E2 genes, which were present in our isolates but were non-existent during the period of 2010–2016. Furthermore, re-emergence of Arg amino acid in place of stop codon in nsP3 gene and Thr at E2:312 positions was observed after 2011. We also used computational tools to assess the effect of the identified mutations on the T cell and B cell epitopes that could influence the protective immune response against this infection.
Chikungunya virus (CHIKV) is a vector-borne alphavirus belonging to the Togaviridae family. It is spread by Aedes mosquitoes and is responsible for acute-onset febrile episodes accompanied by significant musculoskeletal manifestations. Since the first report of CHIKV from Tanzania in 1952 [1], there has been a series of outbreaks throughout the globe including East African, West African and Asian countries [2, 3]. In India, CHIKV was first reported in Kolkata in 1963, followed by epidemics in Andhra Pradesh, Maharashtra and Tamil Nadu in 1964–65 and Barsi (Maharashtra) in 1973 [4]. After a gap of 32 years, CHIKV re-emerged as a massive outbreak in 2006 [5]. Notably, the genotype responsible for this outbreak was of East/Central/South African lineage (ECSA) in contrast to the Asian genotype which had caused the previous outbreaks between 1963 and 1973 [4]. Though the disease remained endemic in India after the outbreak of 2006–2007, there has been a massive upsurge with more than 3-fold rise in confirmed CHIKV cases from 2015 (12% of suspects) to 2016 (41% of suspects) (https://nvbdcp.gov.in). A similar trend was seen in central India where cases rose sharply from 2015 (16% of suspects) to 2016 (37% of suspects) (https://nvbdcp.gov.in). Previous outbreaks of CHIKV have been associated with the emergence of specific mutations in the CHIKV genome. For example, A226V mutation in the E1 protein was found to be responsible for viral adaptation to Aedes albopictus mosquitoes during the 2006–2007 outbreak [6]. Similarly, K211E in the E1 protein and V264A in the E2 protein have been associated with increased adaptation of the virus in the Aedes aegyptii vector during outbreaks in 2009–2010 [7]. In order to explore whether the emergence of any such mutation may have contributed to the 2016–2017 outbreak, we have previously performed a preliminary analysis of partial E1 gene sequences and reported various novel and reemerging mutations [8]. This study expands upon the aforementioned analysis by investigating the entire genome of the 2016–17 outbreak strains, in order to identify novel and re-emerging mutations associated with this outbreak. Furthermore, with the availability of whole genome sequences we explored the impact of these mutations on the antigenic repertoire of the virus.
This study has been carried out in Regional Virology laboratory of All India Institute of Medical Sciences, Bhopal, India, which is a tertiary care teaching hospital. The CHIKV suspected cases attending the outpatient department were tested using Chikungunya IgM capture ELISA kit (supplied by National Institute of Virology, Pune, India) as per the protocol provided by the kit manufacturers. IgM ELISA positive samples were delinked from personal identifiers and further confirmed using RT-PCR based viral detection. The RNA extraction was performed using QIAamp viral RNA mini kit (Qiagen, Germany). The PCR was performed using the SS III platinum one step RT-PCR kit (Invitrogen, US) with CHIKV specific primers (CHIK15F and CHIK16R) as previously described [9]. RT-PCR confirmed cases of CHIKV were further utilized for sequencing. A total of 21 samples were randomly selected for whole genome sequencing of CHIKV, among which 4 isolates (RVL-AIIMS-CH01 to RVL-AIIMS-CH04) were from 2016 and 17 samples were from 2017 (RVL-AIIMSBPL-CH05 to RVL-AIIMSBPL-CH21). The samples were collected between the months of August to October of the respective year, coinciding with the seasonality of Chikungunya cases in our region.
Viral RNA was extracted from 200 μl of serum using the QIAamp viral RNA mini kit (Qiagen, Germany). Three μL of RNA was used as input for cDNA synthesis, performed using Superscript III reverse transcriptase (Thermo Fisher Scientific, USA) and 50 ng of random hexamers (Thermo Fisher Scientific, USA) in a 20 μL reaction, as per the enzyme manufacturer's instructions. Two μL of cDNA was used as input for subsequent multiplex-PCR amplification of the CHIKV genome performed using a tiled-amplicon approach [10]. Barcoded, Nanopore sequencing libraries were prepared according to Oxford Nanopore Technologies' (ONT) 1D Native Barcoding Protocol (available from https://community.nanoporetech.com/protocols), using ONT Native Barcoding kits (EXP-NBD104 and EXP-NBD114) and the ONT Ligation Sequencing kit (SQK-LSK109). The resulting barcoded libraries were loaded onto r9.4.1 flow-cells (FLO-MIN106) and sequenced using MinKNOW software v1.13.1. Base-calling of raw FAST5 files was performed using Guppy v3.1.5 with default settings, discarding reads with a q-score below 7. Base-called reads were de-multiplexed using Qcat v1.0.7 with default settings and individual FASTQ files were aligned to the CHIKV RefSeq sequence (NC_004162.2) using BWA mem v0.7.17 (option-x ont2d) [11]. Primer sequences were trimmed from the aligned reads and draft consensus genome sequences were called using a simple pileup method as previously described [12], masking regions with a read depth <20X.Draft consensus genomes were aligned using MAFFT v7.427 [13] and visualised using AliView v1.2.6 [14], revealing eight short (200–400 bp) regions with insufficient coverage across multiple samples. Additional PCR primer sets were therefore designed to amplify these regions. The resulting amplicons were again barcoded, sequenced, base-called and de-multiplexed in the same fashion described above. The two FASTQ datasets were combined and draft consensus sequences were reconstructed using the same method described above. Base-called FASTQ reads were then realigned to the draft consensus sequences, and primer sequences were again clipped prior to consensus correction, which was performed using Nanopolish variants v 0.11.1 [15](options –ploidy 1 –min-flanking-sequence 10). Variant calls were filtered to include only those with a quality score of > 200 and a support fraction > 70% before being applied to generate the final consensus genomes. Lastly, polished genome sequences were re-aligned using MAFFT v7.427 and visualised using AliView v1.2.6 [14]. Major discrepancies such as indels or multiple consecutive SNPs were manually verified or corrected by referring back to the BAM read alignment file viewed in Tablet v1.19.09.03 [16].
The full length genomes were annotated using VIGOR software available at VIPR server (https://www.viprbrc.org/brc/vigorAnnotator.spg?method=ShowCleanInputPage&decorator=toga) [17]. For phylogenetic analysis, one representative genome from each of 7 lineages [18][ KY703988 (American Lineage), MF773562 (Asian Urban Lineage), GU301781 (Indian Ocean Lineage), EU564334 (Eastern African Lineage), MG649978 (South American Lineage), KP003813 (Middle African Lineage), HM045792 (African/Asian Lineages)and one representative genome each from major countries with reports of CHIKV outbreaks between 2010 and 2015 were downloaded from VIPR database. Similarly, for mutational analysis, all complete genome sequences of Indian origin (n = 30), were retrieved from VIPR database. Phylogenetic analysis was performed with MEGA-X software by using the Maximum Likelihood method with bootstrap support (1000 replicates) and Tamura-Nei model [19]. The tree was drawn to scale, with branch lengths measured in the number of substitutions per site.
In order to assess the modulation in immunological behavior due to sequence variation, the protein sequences from complete Indian CHIKV genomes (from the year 2010–2016) were predicted using in silico prediction algorithms. The immune epitope prediction was performed for B cells, Cytotoxic T cells and T-helper cells.
The prediction of B cell epitopes was performed using LBtope server [20]. LBtope is a Support Vector Machine-based prediction server for B cell epitopes. The window length was set at default value of 15. Residues with probability of correct prediction >60% (default value) were considered as B cell epitopes.
HLA-I binding was predicted using IEDB server [21]. The prediction was performed for all 27 reference HLA-I alleles available at the server. The IEDB HLA-I prediction is based on Artificial Neural Network. The window length was set to 9 amino acids. The epitopes having IC50 < 50nM, were considered as strong binders and those having IC50 between 50 and 500 nM, were considered as intermediate binders. Patterns where IC50 was greater than 500 nM were assigned as non-binders as per the instruction given on the server.
Similar to the HLA-I binding prediction, HLA-II binders were predicted using IEDB server [21] which is based on a position specific weight matrix. The analysis was performed on default window length of 15 amino acids. The reference set of HLA-II alleles (27 alleles) were selected from the server for analysis. Similar to the HLA-I binding prediction, HLA-II binders with IC50 < 50nM, were considered as strong binders and those having IC50 between 50 and 500 nM, were considered as intermediate binders. Non-binders were those which had IC50 > 500 nM. The selection criteria for the reference HLA alleles by the authors in IEDB prediction server includes: 1) the most common specificities in the general population, based on data available from DbMHC and allelefrequencies.net 2) representative of commonly shared binding specificities (i.e., supertypes). In terms of population coverage, the reference sets for class I and II should provide >97% and >99%, respectively.
The phylogenetic analysis performed using representative genome sequences of known lineages and from countries with recent CHIKV outbreaks (since 2010) revealed that the CHIKV isolates from central India during the outbreak of 2016–2017 belonged to ECSA genotype (Figure 1). We had earlier reported on the phylogenetic relationship of CHIKV strains isolated from our region based on partial sequence of the E1 gene [8]. While we observed the phylogenetic proximity of these isolates to recent New Delhi strains of 2015, we wanted to validate these findings based on the whole genome sequences of the CHIKV isolates. The phylogenetic analysis with whole genome sequences revealed that our isolates had closest proximity with strain from Singapore circulating during 2015 and Indian strains of 2014 and 2015 (Figure 1). Furthermore, our isolates also displayed close clustering with strains from Singapore isolated during 2012 and 2013 and from India isolated during 2010, 2012, 2014 and 2015. However, some Indian CHIKV strains, which were also collected in 2016, were found in a separate cluster. In view of the separate phylogenetic clustering of our isolates from majority of Indian isolates sequenced post-2010, we were interested in identifying the specific mutations that had emerged in the viral genome in this geographical region since 2010. Multiple sequence alignment analysis of polyprotein sequences reported from 2010 to 2016 was performed for each position in the protein sequences and four kinds of variations were observed (Table 1). Interestingly, 2 mutations were “novel” as they were non-existent in the intervening period between 2010 and 2016 and were identified in most of our isolates from 2016 to 2017 outbreak. One of these 2 mutations was located in C protein (N79S) and the other was present on E2 protein (A76T). E2:A76T was present in 16 (all from 2017) out of 21 and C:N79S was present in 14 (all from 2017) out of 21 isolates. Interestingly, both of the novel mutations (C:N79S & E2:A76T) were already present in the Singapore strain (2015) but absent in New Delhi strain (2015) (data not shown) (Figure 1). In addition to these 2 novel mutations, E2:M312Tand nsP3:∗524R “reemerged” in our isolates while they were absent in strains reported between 2012 and 2015. These two mutations had previously been observed in 2011 from Kolkata in the eastern part of India. The E2:M312T mutation was present in 7 out of 21 isolates among which 1 isolate was from 2016 and 6 were from 2017. However, nsP3:∗524R was present in all of our isolates. Thirdly 5 additional mutations (E1:I317V, nsP2:H130Y, nsP2:E145D, nsP4:S55N and nsP4:R85G) were found to have emerged in 2015 and continue through 2016–2017 outbreak as “stable” mutations. Apart from these novel, re-emerging (n = 2 each) and stable (n = 5) mutations, we observed 31 locations where the mutations appeared in circulating CHIKV strains in 2015 but were not detected in strains from the 2016–2017 outbreak. These “disappearing” or “unstable” mutations included, 2 positions in C protein (11, 82), 3 positions in E1 protein (31, 55 and 104), 1 position in E2 protein (284), 1 position in nsP1 protein (288), 6 positions in nsP2 protein (130, 145, 149, 157, 213 and 375), 14 positions in nsP3 protein (29, 31, 36, 38, 54, 55, 217, 326, 331, 337, 382, 445, 468 and 470) and 4 positions in nsP4 protein (55, 85, 424 and 555).
To appreciate the immunological relevance of these mutations, we undertook a comparison of predicted B cell epitopes across the CHIKV genome. The 2 novel mutations were not found to alter the humoral antigenicity of corresponding proteins (Supplementary Table S1). However, there were three instances where B cell epitopes disappeared as compared to 2015. This loss of B cell antigenicity was observed in nsP2:375, nsP3:29 and nsP3:426. While the reduction in B cell antigenicity at nsP2:375, nsP3:29 could be attributed to the unstable mutations which appeared at these positions during 2015, the disappearance of B cell epitope at nsP3:426 could be because of a change in the neighboring residues. The B cell antigenicity of all the notable mutations has been displayed in Figure 2A. We did not observe any novel B cell epitopes which appeared due to mutation in 2016–2017. However, there were certain amino acid residues such as E1:211 and E2:312 which were stable as B cell epitopes from 2010 till 2017. Apart from the B cell antigenicity of notable mutations, we also investigated all the B cell epitopic residues. We found 25 such locations which were high confidence B cell epitopes. The epitopic residues and their scores are displayed in Figure 2B.
The protein sequences were predicted for binding affinity to the reference panel of 27 HLA alleles catalogued in IEDB database. We observed two instances (E2:72; nsP3:471) where the HLA binding affinity for allele HLA-A∗30:01 (for E2:72) and HLA-A∗23:01 & HLA-A∗24:02 (for nsP3:471) were found to be reduced from high to intermediate affinity in our isolates (Supplementary Table S2). While no transition from strong to weak HLA I binding was observed in any of the strains, strong HLA I binding was found to be retained at 8 positions among all variable sites across all the strains sequenced between 2010 and 2016. Four of these 8 positions belonged to the nsP4 protein [nsP4:55 (HLA-A∗02:03), nsP4:75 (HLA-A∗31:01& HLA-A∗33:01), nsP4:514 (HLA-A∗02:03) and nsP4:578 (HLA-B∗40:01)]. Similarly, two of these positions belonged to nsP3 protein [nsP3:238 (HLA-A∗02:03), nsP3:352 (HLA-A∗02:01 &HLA-A∗02:06)], one position belonged to C protein [C:23 (HLA-A∗68:02] and one position belonged to nsP2 protein [nsP2:539 (HLA-B∗15:01)]. Beside the HLA I binding affinity of novel, stable, re-emerging and disappearing mutations, we also observed all the strong HLA I bind in whole proteome of our strains. We found 11 such unique HLA I binders. The list of these mutation, their location and the HLA I binding affinity of associated epitope is mentioned in Figure 2C.
We observed 12 unique protein positions which were predicted to be strong HLA II binders among all variable sites, in strains reported from the period 2010 to 2016 as well as in our isolates belonging to 2016–2017. These 12 positions included 6K:54 (HLA-DRB1∗01:01), C:27 (HLA-DRB5∗01:01), E1:291 (HLA-DRB1∗04:05), E1:322 (HLA-DRB5∗01:01), E2:160 (HLA-DRB1∗01:01), E2:375 (HLA-DPA1∗02:01/DPB1∗01:01, HLA-DPA1∗01:03/DPB1∗02:01, HLA-DRB1∗07:01, HLA-DPA1∗03:01/DPB1∗04:02), E2:377 (HLA-DRB1∗01:01, HLA-DPA1∗02:01/DPB1∗01:01), E3:8 (HLA-DRB1∗01:01), nsP2:375 (HLA-DRB1∗01:01), nsP2:643 (HLA-DRB5∗01:01), nsP3:126 (HLA-DRB1∗04:05) and nsP4:500 (HLA-DRB1∗01:01). A decrease in HLA II binding affinity was observed for 7 epitopes, where binding affinity changed from intermediate to weak. To enumerate, C:11 (HLA-DRB1∗11:01), E2:76 (HLA-DQA1∗05:01/DQB1∗03:01), E2:312 (HLA-DRB1∗01:01), nsP2:374 (HLA-DRB1∗04:05), nsP3:217 (HLA-DRB1∗11:01), nsP3:352 (HLA-DRB1∗07:01) and nsP3:468 (HLA-DRB1∗01:01). Notably, E2:76 was novel and E2:312 reemerged in our isolates (2016–2017 outbreak). Furthermore, there were 7 positions where HLA binding affinity altered from weak to intermediate: nsP3:217 (HLA-DRB1∗11:01), nsP3:326 (HLA-DRB1∗04:01), nsP3:331 (HLA-DQA1∗05:01/DQB1∗03:01), nsP3:341 (HLA-DRB1∗01:01, HLA-DRB4∗01:01, HLA-DRB1∗07:01), nsP3:468 (HLA-DRB1∗01:01), nsP4:85 (HLA-DRB1∗13:02) and nsP4:555 (HLA-DRB4∗01:01) (Supplementary Table S3). Similar to HLA I binders, we also catalogued all the strong HLA II binders from all proteins of our strains as mentioned in Figure 2D. We observed 16 such locations which were predicted to be strong HLA II binders. The E1:I317V mutation, which was reported to be unique mutation from central India in our previous study [8], was predicted to lie in a non-B cell epitope during previous years and remained so in 2016–2017. Similar to B cell epitope, the HLA I binding affinity remained unchanged (weak binder) across the years of study including 2016–2017, which could be instrumental in evasion of CD8+ T cell mediated cytotoxicity. However, in case of HLA II binding, we observed that although the HLA II binding affinity for HLA-DQA1∗05:01/DQB1∗03:01 remained unchanged throughout the years of study (strong binder), mutation E1:I317V was found to be intermediate binder for HLA-DRB1∗01:01, HLA-DRB1∗08:02, HLA-DRB1∗11:01 and HLA-DRB5∗01:01; and thus could also be associated with evasion of CD4+ T helper cells in individuals possessing the respective HLA II alleles.
In this paper, we report that the central Indian CHIKV outbreak of 2016–2017 could be attributed to strains belonging to the ECSA genotype, which demonstrated maximum phylogenetic proximity to a Singapore strain and a New Delhi strain, both reported in 2015. More specifically, our isolates displayed two novel amino acid mutations, which were present in the Singapore strain but absent from previously reported Indian strains, suggesting that transmission of CHIKV between Singapore and India may have occurred. We observed a total of four mutations that had emerged in the 2016–2017 outbreak strains compared to the strains reported from 2010 to 2016. These mutations were distributed across E2 (2), C (1) and nsP3 (1) proteins. We attempted to explore the immunological impact of these mutations and observed a number of them that could reduce the affinity of T cell and B cell epitopes and thus enable the virus to subjugate the protective immune response to this infection. India has been affected continually by CHIKV outbreaks since its first outbreak in 1963. Among the major outbreaks, CHIKV reemerged in 1973, 2005 and 2010 and caused heavy health and economic losses to the country [4, 5]. Against a backdrop of sporadic cases occurring each year since 2010, a massive outbreak was observed in 2016 which continued until 2017 [22]. In this study, we observed many novel mutations as well as several that have been previously implicated with survival of virus in the host/vector. For example, mutations such as K211E in E1 protein and V264A in E2 protein were present in the sequenced genomes which have already been associated with increased adaptation of the virus in the Aedes aegyptii vector [7]. Such dual mutation strains were first observed in 2009 in Kerala (India) and continued to be present in outbreaks almost every year since then [23]. Furthermore, the genomes sequenced in this study were found to have codon for Arg in place of opal stop codon in nSP3 gene which is known to attenuate CHIKV virus induced arthritis and pathology. Notably, A226V mutation in E1 protein which is responsible for viral transmission through Aedes albopictus mosquitoes, in 2006–2007 outbreak, was not present in genomes of our isolates. In the phylogenetic analysis, the strains are clustered into two groups where 2016 samples are clustered in one group and most of the 2017 samples clustered in other. Since the chikungunya outbreak studied in the paper ranged from the end of 2016 till starting of 2017 as a single outbreak, the reason for two subgroups in phylogeny could be the genetic drift and filtering of the mutation during the course of an outbreak where the landscape of mutations in the beginning of viral outbreak differs slightly from the end of outbreak. In the past, there have been studies to identify the emerging mutations and their impact on efficacy of potential vaccine candidates in CHIKV strains. However, to the best of our knowledge, there is no study in which the prospect of a CHIKV outbreak has been analyzed in the light of changing antigenicity of the virus. The present study on whole genome sequencing of CHIKV strains builds upon our previous study, which was limited to the partial sequencing of E1 gene during the same outbreak. Our previous study, based on Sanger sequencing, reported the presence of K211E, M269V, D284E, I317V and V322A mutations which were notable because many of these emerged earlier in New Delhi strain of 2010. The whole genome sequencing analysis, undertaken in the current study, demonstrated the same mutations in the E1 gene and thus corroborated our earlier observations. Apart from improving the phylogenetic resolution, the approach of whole genome sequencing enabled us to identify mutations in genes outside of the envelope that may also play a role in virus-host interactions. Our proposed hypothesis, which was based on the emergence of immunomodulatory mutations in this outbreak, is supported by the identification of a range of mutations in several structural and non-structural proteins. Several of these mutations were found to have modest down-regulatory influence on the antigenicity of B-cell and T-cell epitopes and thus collectively evade the protective immune response to the virus and contribute to the massive increase in the number of cases during the 2016–17 outbreak. However, the possible effect of these mutations on improved adaptation to local vector populations also needs to be explored in future entomological studies. Though the signature mutation associated with transmission by Aedes albopictus vector (E1: A226V) was not observed in our study, the role of alternative mutations accounting for vector adaptation can be established only through entomological characterization of locally circulating mosquito populations. Our study suffered from three limitations. Firstly, the sequencing platform used for sequencing the samples was Oxford Nanopore where error rates are relatively higher. However, it has been shown that when coverage depth is good (i.e. >20×), consensus sequences are generally >99.5% accurate. Sequence correction using the nanopolish tool (as we have used) can also help to improve accuracy further (up to ∼99.99%). So the error rate in these genomes should be minimal but they are, of course, still less reliable than those produced using highly accurate technologies such as Sanger dideoxy sequencing or Illumina. The primer sets used for genome amplification were also designed based on the Asian/American genotype, which may account for low genome coverage in some variable regions. However, additional primer sets were designed to specifically improve coverage of these regions. Secondly, since the assessment of the immunogenicity of CHIKV sequences in terms of B cell epitope and HLA binders, was performed using in silico prediction servers, there is a need to confirm the impact of these mutations in suitable animal models and ex vivo studies. Thirdly, the representation of samples from 2016 was relatively less in this study, owing to the non-availability of adequate samples from the initial months of the outbreak that could meet the quality requirements of the NGS experiment.
In this study we have investigated the CHIKV outbreak of 2016–2017 by performing whole genome sequencing of viral samples from central India and analyzed the mutational and immune epitope profile of the viral strains associated with this outbreak with strains from previous years. We report here several novel mutations in structural and nonstructural proteins of CHIKV and also report their potential impact on the occurrence of the outbreak by evading critical B cell and T cell-mediated protective immune responses. The use of the highly accessible and low-cost Nanopore sequencing technology underscores the potential of adopting the same for widespread surveillance of viral pathogens of significant public health interest.
Sudheer Gupta: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper. Ashvini Yadav: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Wrote the paper. Sam Stubbs: Conceived and designed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper. Simon Frost: Conceived and designed the experiments; Contributed reagents, materials, analysis tools or data; Wrote the paper. Kudsia Ansari, Ram Kumar Nema: Performed the experiments; Wrote the paper. Shashwati Nema: Analyzed and interpreted the data; Wrote the paper. Debasis Biswas: Conceived and designed the experiments; Analyzed and interpreted the data; Wrote the paper.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Data will be made available on request.
The authors declare no conflict of interest.
No additional information is available for this paper. |
|
PMC9647482 | 36350652 | Dustin D. Flannery,Sigrid Gouma,Miren B. Dhudasia,Sagori Mukhopadhyay,Madeline R. Pfeifer,Emily C. Woodford,Sara M. Briker,Jourdan E. Triebwasser,Jeffrey S. Gerber,Jeffrey S. Morris,Madison E. Weirick,Christopher M. McAllister,Scott E. Hensley,Karen M. Puopolo | Comparison of Maternal and Neonatal Antibody Levels After COVID-19 Vaccination vs SARS-CoV-2 Infection Comparison of Maternal and Neonatal Antibody Levels After Vaccination vs Infection | 09-11-2022 | This cohort study examines the response to COVID-19 vaccination, exposure to SARS-CoV-2 infection, timing of infection or vaccination, and placental antibody transfers among pregnant persons and their newborns. | Comparison of Maternal and Neonatal Antibody Levels After COVID-19 Vaccination vs SARS-CoV-2 Infection Comparison of Maternal and Neonatal Antibody Levels After Vaccination vs Infection
This cohort study examines the response to COVID-19 vaccination, exposure to SARS-CoV-2 infection, timing of infection or vaccination, and placental antibody transfers among pregnant persons and their newborns.
Pregnant persons are at an increased risk of severe COVID-19 caused by SARS-CoV-2 infection. Pregnancy is associated with an increased risk of mechanical ventilation, intensive care unit admission, and death from COVID-19. COVID-19 during pregnancy may also be a factor in increased risk of stillbirth and complications, such as preeclampsia, preterm birth, and neonatal intensive care unit admission. Although newborns appear to be at a lower risk of severe COVID-19, there are reports of serious neonatal infection and attributable death, and infants are at risk of hospitalization related to SARS-CoV-2 infection. COVID-19 vaccines became available in the US in December 2020. Pregnant persons were excluded from early clinical trials, resulting in uncertainty regarding vaccine administration during pregnancy. With evolving evidence demonstrating both vaccine safety in pregnancy and increased risk of severe infection during pregnancy, the Centers for Disease Control and Prevention released an urgent health advisory in September 2021 that strongly recommended COVID-19 vaccination for pregnant persons. Recent evidence suggests that COVID-19 vaccines are immunogenic and effective in pregnant persons and that maternally derived antibodies can be transferred across the placenta to the newborn after vaccination during pregnancy, as observed after SARS-CoV-2 infection. Most studies to date have been limited by small numbers of vaccinated persons, qualitative antibody assays, exposure to a single vaccine type, or self-report of vaccination. In the midst of the global COVID-19 pandemic, vaccination at any time with any available vaccine is recommended to acutely protect pregnant persons from the disease. As the COVID-19 pandemic evolves, optimal use of available vaccines will be informed by comparative data on vaccine response among pregnant persons and by detailed understanding of vaccination timing to ensure maximal placental antibody transfer. In this study, we leveraged a large cohort of maternal and cord blood serum samples that were tested for antibodies to SARS-CoV-2. The objective was to ascertain the association of vaccine type, time from vaccination, gestational age at delivery, and pregnancy complications with placental transfer of antibodies to SARS-CoV-2.
This cohort study was conducted at Pennsylvania Hospital in Philadelphia, Pennsylvania. Pregnant persons who gave birth at the study site between August 9, 2020, and April 25, 2021, and their newborns were included. The institutional review board of the University of Pennsylvania approved this study and waived the informed consent requirement because the study posed minimal risk and could not practicably be performed without waiver of consent. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Demographic and clinical data, including timing of SARS-CoV-2 exposures and symptoms, SARS-CoV-2 nasopharyngeal polymerase chain reaction (PCR) test results, and vaccine type (messenger RNA [mRNA] vaccine BNT162b2 [Pfizer/BioNTech] and mRNA-1273 [Moderna]; adenovirus vector vaccine Ad26.COV2.S [Johnson & Johnson]) and dates of administration, were collected from review of electronic medical records. Only the first neonates from multiple-gestation deliveries were included in all analyses. Maternal-newborn dyads with incomplete medical records were excluded. Race and ethnicity were self-reported on hospital admission; these data were included given the known disparities in SARS-CoV-2 infection and COVID-19 vaccination. Prepregnancy body mass index (calculated as weight in kilograms divided by height in meters squared) from the first prenatal visit was abstracted from the medical record or from the patient’s self-reported entry in birth registration. International Statistical Classification of Diseases and Related Health Problems, Tenth Revision diagnosis codes were validated and used to identify hypertensive disorders and diabetes, as previously described. Preterm delivery was defined as less than 37 weeks’ gestation, and term delivery was defined as 37 weeks’ gestation or later. During the study period, pregnant persons were routinely screened for SARS-CoV-2 using nasopharyngeal PCR testing when admitted to the hospital for childbirth; testing could also be performed before the pregnancy or earlier during pregnancy and outside of the health system. Persons with medical record report of SARS-CoV-2 symptoms and confirmatory positive result from nasopharyngeal PCR testing were considered to have a symptomatic infection. Symptomatic illness was defined according to definitions provided by the National Institutes of Health. Persons with antibodies to SARS-CoV-2 but without a record of symptomatic illness or COVID-19 vaccination were considered to have asymptomatic infection whether or not they had a positive nasopharyngeal PCR test result.
Pregnant persons routinely have blood drawn for rapid plasma reagin testing when admitted to the hospital for childbirth, and cord blood is routinely collected for newborn blood type and direct antiglobulin testing. Collection of residual serum samples from these specimens was performed as previously described. Serum samples were fully deidentified before antibody level measurements; when results were available, persons who were seropositive were reidentified for manual medical record review by one of us (K.M.P.). The IgG and IgM antibodies to the receptor-binding domain of the SARS-CoV-2 spike protein were measured using enzyme-linked immunosorbent assay; this quantitative assay has been previously described. Serum samples with IgG and/or IgM concentrations of more than 0.48 arbitrary U/mL were considered to be seropositive; concentrations below this cutoff were considered to be seronegative and were assigned a value of 0.24 arbitrary U/mL for statistical analysis.
The demographic, clinical, and antibody characteristics of pregnant persons and newborns were compared according to the following 3 exposures: (1) asymptomatic infection, (2) symptomatic infection, and (3) COVID-19 vaccination with or without infection. Transfer ratios were calculated as newborn IgG concentration divided by maternal IgG concentration. Antibody concentrations and transfer ratios were reported as geometric mean concentrations with 95% CIs and were log2-transformed for statistical analyses. We used scatter diagrams and Spearman rank correlation coefficients to assess the associations between transfer ratio and duration from onset of symptoms or the first positive PCR test result among persons with symptomatic infection or between first vaccine dose and delivery. Mann-Whitney test was used to compare the duration from onset of symptoms or the first positive PCR test result or first vaccine dose to delivery, χ2 test was used to compare seropositivity, and an unpaired, 2-tailed t test was used to compare antibody concentrations and transfer ratios between the analytic groups. We constructed linear regression models to explore the associations between log2-transformed transfer ratio and time from infection or vaccination to delivery, gestational age at delivery, and maternal factors (hypertensive disorders, diabetes, and obesity) that may change placental function. Two-sided P < .05 was considered to be statistically significant. Stata, version 16 (StataCorp LLC) and Prism, version 9 (GraphPad) were used for statistical analyses.
The study cohort consisted of 585 maternal-newborn dyads, with childbirth occurring at a median (IQR) maternal age of 31 (26-35) years and at a median (IQR) gestational age of 39 (38-40) weeks; 31 neonates (5.3%) were born at less than 37 weeks’ gestation (Table 1). The cohort was derived from the 3381 pregnant persons who gave birth during the study period, among whom matched maternal and cord blood serum samples were available for 3119 maternal-newborn dyads (92.3%) (Figure 1). Antibodies to SARS-CoV-2 were detected in 604 pregnant persons (19.4%): 18 (3.0%) had IgM only, 380 (62.9%) had IgG only, and 206 (34.1%) had both IgG and IgM. Because IgM was not expected to cross the placenta, further analyses were restricted to the 585 dyads with maternal IgG and available data (Figure 1). Among these 585 dyads, IgG was detected in cord blood from 557 newborns (95.2%). Of the 28 persons who were seropositive with newborns who were seronegative, the geometric mean IgG concentrations were lower compared with the geometric mean IgG concentrations of persons paired with newborns who were seropositive (1.21 [95% CI, 0.90-1.61] arbitrary U/mL vs 6.46 [95% CI, 5.63-7.42] arbitrary U/mL; P < .001). Demographic characteristics of the study cohort are shown in Table 1. Patients self-identified as being of Asian (35 [6.0%]), Hispanic (117 [20.0%]), non-Hispanic Black (177 [30.3%]), non-Hispanic White (249 [42.6%]), or other (7 [1.2%], including mixed, other, and unknown) race and ethnicity. Of the 585 dyads, 265 (45.3%) had asymptomatic infection and 143 (24.4%) had symptomatic infection. At least 1 dose of a COVID-19 vaccine was administered before delivery in 177 pregnant persons (30.3%): 104 received BNT162b2, 60 received mRNA-1273, and 2 received Ad26.COV2.S. In 11 cases (6.2%), vaccine type was not recorded. A second dose of vaccine was administered before delivery to 126 of 164 persons (76.8%) who were known to be vaccinated with an mRNA vaccine. Compared with individuals with SARS-CoV-2 infection, those vaccinated were older, more often of non-Hispanic White race and ethnicity, and more often had prepregnancy body mass index higher than 30.
Antibody concentrations associated with SARS-CoV-2 infection and COVID-19 vaccination are shown in Table 2. To compare the response to infection vs vaccination, we excluded from this analysis 8 vaccinated persons with a history of infection. The IgG concentrations were higher among persons with symptomatic compared with asymptomatic infection (Table 2). The median (IQR) duration between first vaccine dose and delivery was 42 (26-63) days and was not different between recipients of the BNT162b2 and mRNA-1273 vaccines (Table 2). Two doses of vaccine were administered before delivery to 79% of BNT162b2 vaccine recipients and 77% of mRNA-1273 vaccine recipients. As shown in Table 2, maternal and cord IgG concentrations were higher among mRNA-1273 vaccine recipients compared with BNT162b2 vaccine recipients. The geometric mean maternal IgG concentration of the 169 vaccine recipients without infection was significantly higher compared with the geometric mean IgG concentration of the 408 persons with infection (33.88 [95% CI, 27.64-41.53] arbitrary U/mL vs 2.80 [95% CI, 2.50-3.13] arbitrary U/mL; P < .001). Similarly, the geometric mean cord blood IgG concentration of neonates born to vaccine recipients was significantly higher compared with the geometric mean cord blood IgG concentration of the neonates born to persons with infection (Table 2). A plot of maternal IgG concentration and time from symptomatic, PCR test result–confirmed, well-dated infections or first dose of vaccine to delivery (in days) shows that generally higher antibody levels were associated with vaccination compared with infection (eFigure 1 in the Supplement).
The geometric mean transfer ratio among persons with infection was higher than the geometric mean transfer ratio among vaccinated persons (Table 2). However, transfer ratio that was higher than 1.0 was present for 48 of 51 (94.1%) births at 36 weeks’ gestation or later by 8 weeks after vaccination. The geometric mean transfer ratio was similar between symptomatic and asymptomatic pregnant persons with infection, and between BNT162b2 and mRNA-1273 vaccine recipients (Table 2). Placental transfer ratios were lower after vaccination compared with after infection (0.80 [95% CI, 0.68-0.93] vs 1.06 [95% CI, 0.98-1.14]; P < .001) but were similar between the mRNA vaccines (BNT162b2: 0.85 [95% CI, 0.69-1.06]; P = .25; mRNA-1273: 0.70 [95% CI, 0.55-0.90]). The range for onset of infection to delivery among symptomatic persons was 0 to 384 days, and the first dose of vaccine to delivery among all vaccinated (but without infection) persons ranged in duration from 12 to 122 days; among vaccinated persons after 2 vaccine doses, the range was 19 to 122 days. To ascertain the association between time of infection or vaccination and transfer ratio, we plotted the transfer ratio against time (expressed as days) between onset of symptomatic infection or date of vaccination and delivery. Visual inspection of the plot of all 143 persons with onset of symptomatic infection suggested that the association was not linear over the 384-day range (eFigure 2 in the Supplement). Therefore, in a post hoc analysis, we compared 89 persons with onset of symptomatic infection up to 122 days before delivery (Figure 2A), 159 persons with known date of first dose of vaccine (Figure 2B), and 119 persons with known dates for 2 doses of vaccine (Figure 2C). In each case, the transfer ratio increased linearly over time. The study population included neonates born as early as 23 weeks’ gestation; 23 persons with infection and 8 persons who were vaccinated delivered at less than 37 weeks’ gestation. Placental antibody transfer was detectable as early as 26 weeks’ gestation. When comparing maternal-newborn dyads with neonates born at less than 37 weeks’ gestation or at 37 weeks’ gestation or later, no significant differences were found between geometric mean maternal and cord blood IgG concentrations after infection or vaccination or in geometric mean transfer ratio after infection or vaccination (eTable 1 in the Supplement). To further explore the association of gestational age at delivery with time from symptomatic infection or vaccination, we generated heat maps of mean transfer ratios (Figure 3) and conducted linear regression analyses. The distribution of dyads in each group in the heat map (Figure 3) is shown in eFigure 3 in the Supplement. In linear regression models, we observed a significant association between time from infection or vaccination to delivery and transfer ratio on bivariate analysis, and these associations remained significant in models that included gestational age at delivery and maternal hypertensive disorders, diabetes, and obesity (eTable 2 in the Supplement). For example in the bivariate model, for each day increase in time from infection to delivery, transfer ratio increased by 0.02 (95% CI, 0.01-0.02; P < .001), and for each day increase in time from vaccination to delivery, transfer ratio increased by 0.03 (95% CI, 0.03-0.03; P < .001). In contrast, gestational age at delivery was not associated with transfer ratio on bivariate analysis for infection, and this association remained nonsignificant when accounting for the additional factors (eTable 2 in the Supplement).
Among persons who gave birth between August 9, 2020, and April 25, 2021, at Pennsylvania Hospital, IgG antibodies to the SARS-CoV-2 spike protein were present in higher concentrations after vaccination with mRNA vaccines compared with the antibody levels present after symptomatic or asymptomatic infection and were associated with higher cord blood antibody levels after maternal vaccination compared with maternal infection. Placental transfer ratios were slightly lower after maternal vaccination compared with maternal infection. An increase in transfer ratio was found both with longer duration of time between SARS-CoV-2 exposure (via infection or vaccination) and delivery and with increasing gestational age at delivery. However, multivariate modeling that accounted for the duration of time from infection or vaccination to delivery, gestational age at delivery, and maternal pregnancy comorbidities found that the time from infection or vaccination to delivery was the dominant factor associated with placental transfer. We believe this study expands on the evolving data on COVID-19 vaccination during pregnancy by using deidentified sample collection to minimize consent bias, assessing placental antibody transfer among patients with diverse demographic characteristics, and comparing response to asymptomatic and symptomatic SARS-CoV-2 infection with different COVID-19 vaccines at varying durations between first dose and delivery. We found that antibody levels after vaccination with an mRNA vaccine were at least 10-fold higher than the levels after infection. Furthermore, antibody levels were higher after vaccination with the mRNA-1273 vaccine compared with the BNT162b2 vaccine. Previous studies of smaller numbers of pregnant persons also found that the quantitative IgG response to mRNA vaccines had higher maternal and cord blood IgG antibody levels compared with levels after infection, but these studies were unable to compare IgG responses by specific vaccine types. The reasons for higher antibody levels after the mRNA-1273 vaccine were unclear but may be associated with a higher antigen dose with the mRNA-1273 vaccine compared with the BNT162b2 vaccine. A similar proportion of persons received both doses of each mRNA vaccine and received each vaccine type at similar times before delivery. A comparative effectiveness study conducted among US veterans (92.7% of whom were men, with a median age of 67 years) also found that antibody levels were higher after vaccination with the mRNA-1273 vs the BNT162b2 vaccine. The present study did not address the effectiveness of vaccination in preventing infection among pregnant persons or newborns, although a recent study reported on the effectiveness of maternal vaccination in protecting young infants from SARS-CoV-2 infection. As a novel human exposure, SARS-CoV-2 presents an important opportunity to study placental transfer kinetics given the clear initial timing of exposure at different points during pregnancy and the lack of previous immunity in most cases. Placental transfer of antibodies that are present from conception at low levels is distinct from transfer of antibodies that are present from conception and boosted during pregnancy, such as the intent with tetanus toxoid–reduced diphtheria toxoid–acellular pertussis vaccine administration during pregnancy. In addition, IgG subclass, variation in maternal antibody levels at conception; maternal comorbidities, such as HIV infection; and placental dysfunction that may be present in cases of severe fetal growth restriction or maternal preeclampsia have been described as playing a role in transfer efficiency. Several case reports and series to date have assessed maternal antibody response to different COVID-19 vaccines at various gestations and dynamics of placental antibody transfer after vaccination. Beharier et al reported efficient placental antibody transfer, with transfer ratios generally higher than 1 when the first dose of a vaccine was administered more than 14 days before delivery in a cohort of 86 pregnant persons in Israel who were vaccinated with BNT162b2 . Nir et al reported a correlation between maternal serum and cord blood antibody concentrations in a study that compared 64 women who were vaccinated with the BNT162b2 vaccine with 11 parturient women who had COVID-19 during pregnancy. We did not observe a significant difference in transfer ratio between persons with asymptomatic and those with symptomatic infection, consistent with previous initial findings. However, none of the symptomatic persons with infection were critically ill; therefore, we could not establish whether maternal critical illness is a factor in placental transfer. We explored the contributions of time from infection or vaccination to delivery and gestational age at delivery, revealing that each variable had an independent association with placental antibody transfer. We found no difference in maternal IgG level or transfer efficiency for preterm vs term deliveries when accounting for time from vaccination to delivery. Only a small portion of this cohort had documented SARS-CoV-2 infection before the current pregnancy, but transfer ratios were still robust; the longest interval between infection and delivery that we observed was 384 days, with a transfer ratio of 1.2. As expected, transfer ratios were affected by preterm delivery, although we observed transfer as early as 26 weeks’ gestation at delivery and a transfer ratio higher than 0.5 for vaccine-elicited antibodies as early as 29 weeks’ gestation at delivery. Because antibody levels after vaccination were higher than those after infection, cord blood IgG levels among the 8 cases of preterm delivery in vaccinated persons were significantly higher than the levels observed among cases of term delivery after SARS-CoV-2 infection.
The strengths of this study include the large cohort of pregnant persons with infection and vaccination during the initial phases of the COVID-19 pandemic; diversity in race and ethnicity of the population; pregnancy comorbidities and gestational age at delivery; comparison of antibody responses to different vaccine types; and wide range of timing of infection and vaccination relative to delivery, which provided information on both vaccine response and transplacental antibody dynamics. This study also has limitations. To ascertain the history of infection and vaccination status, we relied on electronic medical record review. We cannot rule out the possibility that some persons whom we deemed to be vaccinated and without infection may have had asymptomatic infections or that some persons whom we deemed to have infection and to be unvaccinated after vaccines became available may have had undocumented vaccinations. We were unable to perform assays for antibodies associated with infection (such as nucleocapsid IgG), which could definitively rule out infection among the vaccinated cohort. In this deidentified study of discarded specimens, we were unable to assess the antibody content in breastmilk, duration and durability of newborn antibody, or degree of newborn protection from infection. Furthermore, we did not identify the IgG subclass of vaccine-elicited antibodies.
This cohort study found that concentrations of maternal IgG antibodies to SARS-CoV-2 in pregnant persons after vaccination with mRNA vaccines were higher than the levels after viral infection, but placental antibody transfer ratios were lower after vaccination than after infection. Placental transfer and cord blood IgG concentration were detectable as soon as 15 days after the first dose of an mRNA vaccine, and transfer ratios increased for several weeks after the first vaccine dose. These findings suggest that time from infection or vaccination to delivery was the most important factor in transfer efficiency, and these findings can inform optimal COVID-19 vaccination strategy during pregnancy. |
|
PMC9647491 | Chuan Teng,Fanhua Kong,Jinggang Mo,Weidong Lin,Chong Jin,Kunpeng Wang,Ying Wang | The roles of RNA N6-methyladenosine in esophageal cancer | 05-11-2022 | Esophageal cancer,N6-methyladenosine,m6A,Epigenetics,Immunotherapy,RNA methylation | Esophageal cancer is a malignant tumour with a high degree of malignancy and high mortality. Its pathogenesis and treatment strategy remain unclear. N6-methyladenosine (m6A) is important for various biological functions in RNA modification and is currently being investigated extensively. It plays an essential role in RNA modification. m6A modification is a dynamic process that reversibly regulates the target RNA through its regulatory factors and plays an important role in several diseases, especially cancer. However, the role of m6A in esophageal cancer remains elusive. RNA modification and splicing are regulated by RNA methylation regulators called ‘writers’ (methyltransferases), ‘erasers’ (demethylases) and ‘readers’ (modified RNA-binding proteins). These regulatory factors recognise and bind to RNA methylation sites, regulate biological functions such as RNA splicing and translation and influence the occurrence, development, invasion and metastasis of tumours. Considering the importance of m6A modification, we reviewed the regulatory mechanisms, biological functions and therapeutic prospects of m6A RNA methylation regulators in esophageal cancer. | The roles of RNA N6-methyladenosine in esophageal cancer
Esophageal cancer is a malignant tumour with a high degree of malignancy and high mortality. Its pathogenesis and treatment strategy remain unclear. N6-methyladenosine (m6A) is important for various biological functions in RNA modification and is currently being investigated extensively. It plays an essential role in RNA modification. m6A modification is a dynamic process that reversibly regulates the target RNA through its regulatory factors and plays an important role in several diseases, especially cancer. However, the role of m6A in esophageal cancer remains elusive. RNA modification and splicing are regulated by RNA methylation regulators called ‘writers’ (methyltransferases), ‘erasers’ (demethylases) and ‘readers’ (modified RNA-binding proteins). These regulatory factors recognise and bind to RNA methylation sites, regulate biological functions such as RNA splicing and translation and influence the occurrence, development, invasion and metastasis of tumours. Considering the importance of m6A modification, we reviewed the regulatory mechanisms, biological functions and therapeutic prospects of m6A RNA methylation regulators in esophageal cancer.
Esophageal cancer (EC) is one of the most common malignant tumours worldwide, with extremely high mortality and low survival rates. The most common pathological classifications include oesophageal squamous cell carcinoma (ESCC) and oesophageal adenocarcinoma (EAC) [1, 2]. More than half a million people die of esophageal cancer every year, constituting 5.3% of all cancer deaths worldwide. There are significant differences in the incidence of esophageal cancer cases worldwide [3]. In East Asia, the incidence of ESCC is high, which is mainly related to smoking and other factors. EAC is mostly reported in Western countries, and its incidence is positively correlated with obesity-related diseases and gastroesophageal reflux disease (GERD) [4]. Esophageal cancer is an aggressive cancer with a 5-year survival rate of approximately 18.4% owing to its extremely high mortality rate [1]. ESCC remains the most common type of esophageal cancer; however, its incidence has declined in East Asia and increased in Europe and the United States over the past few decades [5]. Differences in lifestyle and genetic background may be the main reasons for these regional differences. At present, surgical resection remains the mainstay of treatment for esophageal cancer [6]. In the United States, only 18% of patients with esophageal cancer were diagnosed without metastases, and 40% of patients were reported to have distant metastases on diagnosis, indicating a poor prognosis of esophageal cancer [7]. It is projected that there will be 15,000 new EAC cases every year in the United States until 2030 [8]. Therefore, in-depth investigation of the pathogenesis of esophageal cancer is significant to improve the survival rate of patients. Gene transcription abnormalities associated with esophageal cancer, including chromosomal and tumour cell mutations, were a major research focus in the past [9]. Studies have shown that gene mutation is an important mechanism of esophageal cancer, and various genetic mutations are found in patients with esophageal cancer. Of these mutations, TP53 mutation is the most common. High mutation rates of EGFR, CCND1, CDK4/CDK6 and MDM2 genes were reported in ESCC [10]. In addition, CCNE1, cyclin E and MGST1 also have high mutation rates in EAC, and the mutations are closely related to the occurrence and development of esophageal cancer [11]. Furthermore, the upregulation of vascular endothelial growth factor C (VEGF-C) was positively associated with event-free survival, whereas the allele variation of FLT1 increased the risk of death [10, 12]. It is noteworthy that, in addition to epigenetic changes caused by gene mutations, drug resistance owing to chemical drug therapy for esophageal cancer has become increasingly prominent. Studies on the plasticity and chemotherapeutic resistance of esophageal cancer cells indicate that epigenetic modifications are involved in the regulation of abnormal phenotypic changes [13, 14, 15]. Moreover, epigenetic modifications are reversible, compared with genetic defects; therefore, epigenetic inheritance is considered as a prospective research focus. Increasing evidence suggests that epigenetic disorders influence the occurrence and development of esophageal cancer [16, 17, 18]. N6-methyladenosine (m6A), or N6- methylated adenosine, has been extensively investigated [19]. m6A methylation regulators consist of three components, namely, methyltransferases (writers), demethylases (erasers) and binding proteins (readers) [20]. Recent studies have demonstrated that m6A either promotes carcinogenesis or inhibits malignant tumours. Zhang et al. demonstrated that m6A-mediated methylation modification regulates changes in the tumour microenvironment invasion [21, 22]. In addition, Han et al. reported that m6A modification further enhanced YTHDF1-mediated immune regulation. YTHDF1 can be used as a potential therapeutic target for malignant tumours. Moreover, YTHDF1 also plays an important role in tumour immune evasion; however, the specific mechanism requires further investigation. m6A modification is important for the prognosis of patients with cancer [23]. The correlation between the expression of m6A methylation regulators and immune invasion in esophageal cancer requires to be investigated comprehensively. Therefore, we conducted a synthetical analysis of m6A modification and analysed the role of m6A modification-mediated epigenetic changes in esophageal cancer. We further elucidated that m6A methylation regulatory factors may serve as diagnostic markers and therapeutic targets for prospective studies on esophageal cancer.
m6A modification is dynamic and reversible, and the methyltransferase writers mainly include KIAA1429 (VIRMA), METTL3, RBM15, WTAP, ZC3H13, METTL16, METTL14 and CBLL1 [24]. The main feature of METTL3 and METTL14 is that they both contain a methyltransferase domain (S-adenosylmethionine; SAM), which is used to transfer methyl groups to adenosine at N6 [25]. METTL3 catalyses the formation of the methyltransferase complex, and METTL14 induces serine phosphorylation on METTL3 [26]. In addition, METTL14 expression is upregulated in pancreatic cancer cells, and its downregulation can increase cisplatin-induced apoptosis and autophagy in pancreatic cancer cells [20]. METTL16 is another methyltransferase that binds to non-coding RNAs (ncRNAs) to regulate mRNA transcription and maintain SAM homeostasis [27, 28]. KIAA1429 is a crucial element of the methyltransferase complex, and its main function is to guide regionally selective deposition of m6A [29]. Furthermore, it also regulates the expression of sex-lethal genes by alternative splicing (AS) of pre-mRNA with WTAP [30]. ZC3H13 is a typical CCCH zinc finger protein and a key factor in the nuclear localisation of the ZC3H13–WTAP–Virilizer–Hakai complex [31]. RBM15 is an RNA-binding protein. As a member of the SPEN (split ends) protein family, RBM15 is involved in m6A modification and as regulation and exhibits inhibitory functions in multiple signalling pathways [32]. CBLL1 is highly expressed mainly in non-small cell lung cancer cells and promotes tumour proliferation and invasion (Figure 1) [33].
To date, only two m6A demethylases have been identified, namely, ALKBH5 and fat mass and obesity-associated gene (FTO) [34, 35], which belong to the ALKB dioxygenase family, and their biological functions are mainly catalysed by Fe2+ and α-ketoglutaric acid [34]. ALKBH5 is a major m6A demethylase that plays an important biological role in human cancers and non-cancer diseases. For example, it plays a dual regulatory role in various cancers and reproductive system diseases [36]. FTO, as an obesity-susceptibility gene, plays a central role in regulating food intake. In addition, FTO plays a peripheral role by influencing lipolysis in adipose tissues [37]. Furthermore, FTO affects mRNA stability and translation efficiency by regulating m6A modification. In addition to FTO and ALKBH5, other m6A demethylases require to be identified (Figure 1).
m6A-binding proteins mainly include YT521-B homology (YTH) domain-containing proteins (YTHDC1, YTHDC2, YTHDF1, YTHDF2 and YTHDF3), heterogeneous nuclear ribonucleoprotein (hnRNP) family members (hnRNPA2/B1, hnRNPC and hnRNPG), FXR family members, IGF2BP family members, eIF family members and G3BPs family members [26, 38]. The main characteristic of m6A-binding proteins is that they all have a conserved m6A-binding domain and bind to the RRm6ACH sequence [39, 40]. m6A-binding proteins bind to mRNAs that exhibit 10–50 times higher affinity than that exhibited by unmodified mRNA. Moreover, they encode information regarding m6A modification and perform various biological functions by interacting with modified RNA [34]. The hnRNP family is involved in various cellular functions, including the regulation of transcription, mRNA metabolism and translation [41]. hnRNPs represent a large family of RNA-binding proteins (RBPs) that contribute to many aspects of nucleic acid metabolism, including AS, mRNA stabilisation, transcription and translation. Many hnRNPs exhibit common characteristics but differ in domain composition and functional properties [42]. YTH family proteins are the first identified m6A reading proteins, which are involved in the occurrence and development of cancerous tumours by regulating the targeted RNA metabolism, including RNA splicing, RNA output, translation and degradation (Figure 1) [43].
The regulation of post-expression genes is implemented in four processes: transcription, post-transcription, translation and post-translation. m6A regulates RNA transcription and gene expression after RNA transcription by modifying the structure of RNA or by specific binding in the form of binding proteins [44]. The specific mechanism includes the labelling of m6A into the newly formed mRNA after transcription. Furthermore, the post-transcriptional mRNA is recruited to an m6A site or flanking sequence through splicing factors for AS and is then transformed to mature mRNA. Subsequently, mRNAs that contain m6A are recognised by YTHDC1 and transported to the cytoplasm [44]. mRNAs in the tumour cytoplasm are cleaved, matured and positioned on the ribosome for translation. In addition, in the cytoplasm, YTHDF family proteins enhance m6A-mediated mRNA metabolism. YTHDF1 can specifically bind to the m6A site and interact with the initiation factor eIF3 to promote the initiation of translation and protein synthesis [45]. YTHDF3 interacts with YTHDF1 to promote mRNA translation and accelerates the decay of m6A mRNA by interacting with YTHDF2 [46]. Furthermore, METTL3 promotes the translation of m6A-modified mRNAs in the cytoplasm, independent of its methyltransferase activity [47]. Choe et al. reported that METTL3 enhanced translation by interacting with the mRNA near the termination codon and supported the mRNA cyclisation mechanism to achieve ribosome circulation and translation control [47]. In addition, Wang et al. [48] conducted a study with 200 patients with ESCC and reported that METTTL3 was upregulated in tumour tissues. METTTL3 recruited YTHDF for adenomatous polyposis coli (APC)–mRNA degradation by upregulating m6A of the APC gene. The expression of APC was decreased, and the expression of β-catenin, cyclin D1, C-MyC and PKM2 was increased, which promoted aerobic glycolysis, proliferation and metastasis of ESCC cells. Moreover, METTL3 promoted ESCC metastasis by enhancing glutaminase-2 (GLS2) expression. This study was the first to demonstrate that GLS2 is regulated by METTL3 through m6A modification. Therefore, METTL3/GLS2 signalling is expected to be a potential therapeutic target for ESCC anti-metastasis strategies [49]. Hou et al. found that METTL3-mediated AKT signalling pathway can promote the occurrence and development of esophageal cancer, and METTL3 upregulation can promote the proliferation, migration and invasion of ESCC cells and inhibit cell apoptosis [50]. Meanwhile, METTL3 can also increase m6A in EGR1 mRNA in a YTHDF3-dependent manner, enhance its stability, and activate EGR1/Snail signal [51]. Thus, METTL3 is upregulated in esophageal cancer, promotes proliferation and metastasis and is expected to be an independent biomarker for prognosis [52, 53, 54]. m6A writers and erasers are located in the plaques associated with mRNA splicing factors, suggesting that m6A is functionally associated with mRNA splicing [55]. m6A can exert the splicing function of mRNA on pre-mRNA by recruiting hnRNPB2A1 or altering the local structure and can increase the access of splicing factor hnRNPC/hnRNPG, thus affecting transcription [56]. In view of this, Guo et al. [57] used The Cancer Genome Atlas database and found that m6A is significantly enhanced, m6A regulatory factors are significantly upregulated, and the expression of ALKBH5 and hnRNPA2B1 is upregulated in patients with esophageal cancer, thus suggesting that m6A facilitates the prognosis of patients. hnRNPA2B1 upregulation can promote the proliferation and metastasis of esophageal cancer and the expression of the fatty acid synthase adenosine triphosphate citrate lyase (ACLY) and acetyl–coenzyme A carboxylase-1 (ACC1). Studies have demonstrated that ACLY and ACC1 are carcinogenic factors that can promote the proliferation and metastasis of esophageal cancer. Therefore, ALKBH5 and hnRNPA2B1 may serve as diagnostic and prognostic markers and therapeutic targets for esophageal cancer. Nagaki et al. reported that ALKBH5 upregulation was associated with a poor prognosis of esophageal cancer, and the inhibition of ALKBH5 delayed the progression of esophageal cancer cell cycle, resulting in cell stagnation at G0/G1 phase. In ALKBH5-deficient cells, CDKN1A (P21) expression was significantly upregulated, and the knockdown of ALKBH5 increased m6A modification and CDKN1A mRNA stability [58]. FTO is an m6A-modified demethylase that has been associated with various tumours. Liu et al. demonstrated that FTO was upregulated in ESCC tissues and promoted the proliferation and migration of ESCC cells by upregulating MMP13 [59]. FTO can also promote the formation of EC cell lipid droplets by enhancing the expression of HSD17B11, thus promoting EC proliferation, invasion and tumorigenicity [60].By analysing the TCGA database, Xun et al. demonstrated that m6A upregulation resulted in a worse prognosis of patients with ESCC, which may provide important information for diagnosis and treatment strategies [61]. The mechanism of m6A factors that regulate mRNA in esophageal cancer is demonstrated in Figure 2 and Table 1. The study design, key protocols/methods, and materials of m6A regulators in regulating mRNAs as shown in Table 2.
ncRNAs are RNAs that do not code for proteins. They include long ncRNAs (lncRNAs), microRNAs (miRNAs) and circular RNAs (circRNAs), with various known and unknown functions. The common feature of these RNAs is that they can be transcribed from the genome but cannot be translated to proteins; therefore, they perform their biological functions at the RNA level. Studies have indicated that m6A methylation regulatory factors are important for regulating ncRNA expression and function [27]. In this review, we have systematically described the regulatory mechanism of m6A modification in ncRNAs. The mechanism of m6A factors that regulate ncRNA in esophageal cancer is demonstrated in Figure 3 and Table 3. The study design, key protocols/methods, and materials of m6A regulators in regulating ncRNAs as shown in Table 4.
miRNAs refer to a class of endogenous small RNAs with a length of about 20–24 nucleotides, which play an important regulatory role in cells and are involved in gene silencing or post-transcriptional gene expression regulation [62]. Li et al. analysed esophageal cancer data from the TCGA database and found that the expressions of 25 m6A regulators were increased and positively correlated. hnRNPA2B1 upregulation promotes lymph node metastasis of esophageal cancer and is associated with poor prognosis. Moreover, the knockdown of hnRNPA2B1 significantly downregulated miR-17, miR-18a, miR-20a, miR-93 and miR-106b, resulting in decreased proliferation and metastasis of esophageal cancer cells [63]. In addition, Xue et al. demonstrated that miR-193a-3p was upregulated in ESCC tumour tissues, compared with normal tissues, and promoted invasion and metastasis, and a high expression of miR-193a-3p promoted the recurrence of ESCC. In addition, miR-193a-3p and ALKBH5 can regulate each other by forming a closed loop. Therefore, the mutual regulation of miR-193a-3p and ALKBH5 affects the progression of ESCC [64]. In addition, miR-20a-5p expression was up-regulated in esophageal cancer, and METTL3 increased the expression of miR-20a-5p by improving m6A modification. In addition, miR-20a-5p upregulation promotes the invasion and migration of esophageal cancer by targeting NFIC transcription [65]. ALKBH5 has been confirmed to inhibit the development of esophageal cancer mainly through m6A/DGCR8-dependent demethylation of pri-miR-194-2 and inhibition of miR-194-2 biogenesis, followed by inhibition of esophageal cancer progression by ALKBH5/miR-194-2/RAI1 axis [66].
lncRNAs are a group of endogenous RNA molecules defined as ncRNAs with a length of more than 200 nucleotides. They have been reported to play an important role in many diseases, including ESCC [67]. A few studies have investigated the mechanism of m6A modification in lncRNAs. Wu et al. reported downregulation of Y-linked lncRNA LINC00278 in men with ESCC, which encodes a Yin Yang-1 (YY1)-binding peptide named YY1BM. YY1BM is involved in the progression of ESCC and inhibits the interaction between YY1 and androgen receptors, thus reducing the expression of eEF2K through the androgen receptor signalling pathway. YY1BM downregulation significantly upregulated the expression of eEF2K, inhibited apoptosis and enabled ESCC cells to better adapt to nutrient deficiency. Moreover, smoking decreased the m6A-modified LINC00278 and YY1BM translation [68]. In addition to surgical treatment for ESCC, platino-based drugs in combination with 5-fluorouracil (FP) are the first-line treatment for patients with ESCC, especially for those in advanced stages [69]. A growing body of evidence suggests that m6A modification plays an important role in tumour progression and chemoradiotherapeutic resistance [70]. Zhang et al. reported increased m6A levels and abnormal expression of SNHG3/miR-186-5p in patients with ESCC after platinum therapy. SNHG3/miR-186-5p is involved in the regulation of m6A expression level by targeting METTTL3. Therefore, the regulation of m6A level may offer a new strategy to improve the effects of platinum drugs on patients with ESCC [71]. Another study showed that LncRNA LINC00022 was up-regulated in ESCC, and FTO demethylated LncRNA LINC00022 and inhibited its degradation, promoting tumor growth in ESCC [72]. The expression of IGF2BP2, TK1 and LncRNA CCAT2 were upregulated in ESCC cells and tissues, while the expression of miR-200b was inhibited. CCAT2 bound to miR-200b and reduced its expression, resulting in the upregulated expression of IGF2BP2. IGF2BP2 enhances the stability of TK1 mRNA by recognizing the m6A modification of TK1, thereby enhancing its expression and promoting the migration and invasion of ESCC cells [73].
circRNA is a new type of ncRNA, which forms a covalently closed continuous loop through reverse splicing, unlike linear RNA. Recently, m6A modification has been reported to be widespread in circRNAs and has the same read–write mechanism as mRNAs [74]. Wang et al. used plasma samples from 10 patients with ESCC, including patients with different tumour, lymph node and metastasis (TNM) stages, and 5 normal controls to screen the expression profiles of circRNAs and analyse the characteristics of circRNAs. They found that plasma circ-SLC7A5 levels in the patients were associated with TNM staging. In addition, circ-SLC7A5 contained a large number of m6A modification structures and exhibited a high translation potential. Moreover, circ-SLC7A5 exhibited a high affinity for binding to open reading frames and contained more miRNA-recognition elements (MRE), suggesting that circRNAs play a more diverse role in ESCC [75].
Maladjustment of m6A regulators leads to abnormal m6A modification in key transcripts and abnormal regulation of the expression of these cancer-related genes. Therefore, m6A site mutations may play a role in cancer by interfering with m6A deposition [44]. Yang et al. assessed the relationship between m6A modification gene variation and ESCC risk and reported the single nucleotide polymorphism rs2416282 in the promoter region of YTHDC2, which was significantly associated with the occurrence and development of ESCC. Furthermore, rs2416282 regulated YTHDC2 expression, and the knockout of YTHDC2 significantly promoted the proliferation rate of ESCC cells by affecting various cancer-related signalling pathways [76]. As a risk variant of YTHDC2, rs2416282 alters YTHDC2 expression and reduces ESCC risk. Therefore, gene variation plays an important role in the regulation of m6A modification. However, further investigation is required to comprehensively understand the pathogenesis of ESCC. Although m6A modification does not change base pairing and coding, it broadly affects gene expression in multiple levels through interacting with diverse reader proteins and associated complexes. Therefore, dynamic m6A modification is critical for many normal bioprocesses, including self-renewal and differentiation of embryonic stem cells and hematopoietic stem cells, tissue development, circadian rhythm, heat shock or DNA damage response, and sex determination [77]. Recently, abnormal reductions or increases in m6A abundance have been found in some types of cancer, and this disorder may be associated with cancer progression and clinical outcomes. For example, it has been reported that m6A abundance (detected by m6A spot blotting or ELISA-like colorimetry) was significantly increased in mRNA or total RNA in human gastric cancer tissues compared with normal control tissues [78]. It is now understood that abnormal regulation of m6A regulators, particularly writer and erasers, leads to abnormal modification of m6A in key transcripts and to abnormal post-transcriptional regulation of the expression of these cancer-related genes. Therefore, it is reasonable to speculate that m6A mutations in these transcripts may interfere with m6A deposition and thus play a role in cancer. It has been reported that codon NO.273 of TP53 pre-mRNA G>A mutation (resulting in R273H mutation) makes colon cancer cells resistant in an m6A dependent manner [79]. More recently, a large-scale population study identified a missense variant, the rs8100241 variant located in the exon of ANKLE1 with a G > A change (resulting in Ala > Thr change) in CRC, and showed that it was associated with decreased risk of CRC [80]. Mechanically speaking, compared with rs8100241[G] allele, rs8100241[A] allele can be methylated by m6A MTC and recognized by YTHDC1, thereby increasing the protein expression of ANKLE1. ANKLE1 is a potential tumour suppressor that inhibits cell proliferation by maintaining genomic stability [80]. In addition to the acquisition of new m6A sites due to cancer mutations, mutations that lead to loss of m6A modifications may also exist and contribute to cancer development and drug response. Several online tools combine m6A site information with SNP information to facilitate the investigation of functional m6A site mutations in cancer [81].
The immune system is the host's defence against infection and disease. Immunotherapy is a new cancer treatment strategy that has been widely used in the treatment of various solid tumours, including gastrointestinal tumours [82, 83]. In recent years, m6A regulatory factors have been extensively investigated to determine their function in tumour immunotherapy and immune avoidance. Currently, tumour immunotherapy and m6A modification are the most promising therapeutic strategies that are under investigation in studies concerning the pathogenesis and prognosis of esophageal cancer. Liu et al. used the Tumor Immune Estimation Resource (TIMER) database to analyse the relationship between glucose transporter-1 (GLUT1) expression, m6A modification and immune infiltration in esophageal cancer. They found that GLUT1 expression was upregulated in esophageal cancer and was associated with infiltration of various immune cells. For example, when GLUT1 expression was upregulated, the number of memory B cells was reduced. Therefore, upregulated GLUT1 expression in patients with esophageal cancer may trigger an anti-tumour immune response and play an important role in the regulation and recruitment of infiltrating immune cells in esophageal cancer [84]. There has been encouraging progress in immunotherapy for esophageal cancer, and some related immunosuppressants have entered clinical trials and exhibited persistent anti-tumour activity and controllable adverse reactions [85]. m6A modification has been reported to play an important role in tumour immunotherapy and immune avoidance, and therapeutic strategies that target m6A regulators have become the focus of current and future research [23]. Guo et al. analysed the transcriptional sequencing data and clinical information regarding 20 m6A methylation regulators from 453 patients with ESCC and found that METTTL3, WTAP, IGF2BP3, YTHDF1, hnRNPA2B1, hnRNPC and PD-L1 were significantly upregulated in the patients. Immune scores showed a significant increase in the number of CD8+ T cells, stationary mast cells and regulatory T cells (Treg) in the patients. These results indicated that m6A regulators are the key mediators of PD-L1 expression and immune cell infiltration, which is expected to be an important target of ESCC immunotherapy [86]. In addition, Zhao et al. also demonstrated that the abnormal expression of m6A regulators was significantly correlated with the expression of ESCC immunomodulators (immunosuppressants, immunostimulants and major histocompatibility complex [MHC] molecules) and the level of immune infiltration [87]. YTHDF1 recognizes m6A-labeled transcripts encoding lysosomal proteases and increases their translation in dendritic cells. Loss of YTHDF1 promotes cross-expression of tumor antigens and cross-initiation of CD8+ T cells in vivo. In addition, the absence of YTHDF1 enhances the therapeutic effect of PD-L1 checkpoint blockade [23]. In melanoma, increased FTO levels promote tumor growth by decreasing m6A methylation in PD-1 (PDCD1), CXCR4, and SOX10 and preventing YTHDF2-mediated RNA decay. FTO knockout in melanoma cells sensitized tumor cells to interferon γ (IFNγ) in vitro and promoted mouse melanoma response against PD-1 antibodies [88]. METTL3 promotes circIGF2BP3 cycling and protects PD-L1 from proteasome-mediated degradation in a YTHDC1-dependent manner, thereby negatively regulating CD8+ T cell infiltration and promoting immune evasion of non-small cell lung cancer (NSCLC) cells [89]. In addition, m6A modification was positively correlated with the number of CD8+ T cells in pancreatic cancer immune microenvironment, suggesting that m6A modification may regulate CD8+ T cell aggregation [90]. These data suggest that regulation of m6A regulators can be combined with anti-PD-1/PD-L1 blockade to improve anticancer immunotherapy. Therefore, future studies should be focused on the function of m6A methylation regulators in ESCC tumour immunotherapy. By regulating RNA transcription, splicing, processing, translation and decay, m6A modification is involved in the occurrence and metastasis of various malignant tumours. As the modification of m6A and its mechanism are involved in the occurrence, development, maintenance and drug resistance of various types of diseases, the study of m6A regulatory factors in immune response is still in its infancy. m6A gene modification supports rapid phenotypic variation in the disease. In malignant tumors, m6A regulators contribute to changing transcriptional patterns of gene expression in malignant tumors, which may contribute to tumor immune evasion and persistence in other diseases. However, these meaningful conclusions are due to the absence of one of the regulatory factors of m6A modification, while other mRNA or molecular component methylation levels associated with key phenotypes remain to be explored. Currently, most studies that rely on m6A immunophenotypes use cells in which an important component of the m6A regulatory complex has been deleted. Determining which m6A regulatory factors drive the observed phenotypes and whether other regulatory factors co-regulate the results remains to be studied. In the context of immune regulation, how immune signals can be gene-specific for specific m6A regulatory factors. Some m6A-mediated transcripts are immune-related genes that may drive phenotypic changes in the immune system. In addition, m6A modification inhibits the expression of MHC I on the surface of cancer cells, meaning that tumors can avoid inherent and adaptive immune responses. The m6A regulatory factor FTO also upregulates immune checkpoints and promotes immune escape. m6A seems to increase the activity of immunosuppressive cells, suppressing the activity of immune-promoting cells. The same result was found for chemokines. At present, numerous studies have reported the relationship between m6A methylation and TME immune cell infiltration. For example, METTL3-mediated mRNA m6A methylation promotes dendritic cell activation and function, and knockdown of METTL3 in dendritic cells results in impaired phenotype and functional maturation of dendritic cells, decreased expression of costimulatory molecules CD40 and CD80 and cytokine IL-12, and reduced ability to stimulate T cell responses [91]. Yin et al. found that ablation of METTL3 in myeloid cells promoted the infiltration of tumor-associated macrophages and regulatory T cells into tumors [92]. Dong et al. demonstrated that down-regulation of METTL14 expression leads to differentiation of CD8+ T cells along a dysfunctional trajectory, promoting immune escape [93]. In addition, knockdown of ALKBH5 in GBM cells significantly inhibited the recruitment and immunosuppression of tumor-associated macrophages, and the expression and secretion of CXCL8/IL8 were also significantly inhibited [94]. These results suggest that m6A modification may play an active role in macrophage activation and the pathogenesis of various inflammatory diseases [95]. In addition, although much progress has been made in m6A modification, there is still a gap in understanding how m6A modification affects immune response, especially immune avoidance. At present, immunotherapy has become the most popular cancer treatment method, among which targeting immune checkpoints is also closely related to m6A [96]. In the above, we discussed various mechanisms of immune evasion and immune regulation by m6A modification in cancer. Among the many immunotherapy strategies, immune checkpoint blockade has shown remarkable efficacy. Studies have shown that m6A modification is involved in the regulation of PD-1 and PD-L1, and affects the tumor response to anti-PD-1 therapy. For example, in melanoma, METTL3/14, ALKBH5, and FTO can all modulate tumor response to anti-PD-1 therapy [97]. Inhibition of FTO can reduce tumor resistance to PD-1 treatment [88]. In addition to immunotherapy targeting PD-1, m6A modification is also involved in the regulation of cytotoxic T-lymphocyte antigen 4(CTLA-4) targeted therapy [96]. In addition, m6A modification has also been implicated in the regulation of TIM3 (T-cell immunoglobulin-3), TIGIT (T Cell Immunoreceptor With Ig And ITIM Domains) and LAG3(Lymphocyte Activation Gene3) [96]. Although the relationship between these immune checkpoints and m6A has not been thoroughly studied, their potential impact on tumor immunotherapy should not be ignored. These findings suggest that the development of specific inhibitors targeting m6A regulators may be helpful for antitumor therapy. Regulation of m6A is closely related to the development and function of immune cells. Targeted regulation of m6A can enhance the function and infiltration of immune cells in the tumor microenvironment and enhance the tumor response to anti-PD-1 therapy through a variety of signaling pathways. It can be seen that the use of specific inhibitors targeting m6A regulators can combine antitumor activity with antitumor immunity and exert a comprehensive therapeutic effect. This will also be the hot direction of future studies on m6A modification involved in immunotherapy.
Although many scholars have reported studies on m6A modification in tumor therapy, precise targeted therapy of m6A needs to be explored due to the complex mechanism of tumor formation. It affects the expression of a series of downstream oncogenes or transcription factors by changing the m6A level of some specific genes corresponding to mRNA in cells. Regulating the level of m6A in tumor cells may be the entry point for radiotherapy, chemotherapy and drug therapy. So far, more studies have focused on the identification of m6A-modified mRNA targets in cancer, while fewer studies have focused on m6A modification on ncRNAs, which is one of the future research directions. Thus, the treatment of cancer still has a long way to go in the future. Studying how m6A modification affects ncRNA production, cell location, and function, as well as how these processes are related to cancer, will greatly improve our understanding of m6A cell function and the currently unrecognized function of ncRNA. Recently, modifications of m6A have been found on chromosomal associated regulatory RNAs (carRNAs), including promoter associated RNAs, enhancer RNAs, and repeat RNAs, and have been shown to regulate chromatin state and transcription [98]. Whether these carRNA species play a role in cancer remains an interesting topic for further study. In addition to immune checkpoints and cell therapy, some regulatory cells related to tumor immunity also have the potential to be targets [99]. For example, regulatory T cells (Tregs) in the tumor microenvironment inhibit the body's immune response to the tumor, which is a factor hindering immunotherapy. It can change the tumor response to immunotherapy through the regulation of m6A [100]. In addition, knockdown of ALKBH5 reduces Treg infiltration in melanoma and enhances the response to anti-PD-1 therapy [101]. In addition, since m6A modification also plays an important role in mediating tumor responses to chemotherapy, radiotherapy and immunotherapy, targeted therapies targeting m6A regulatory factors can also be combined with chemotherapy, radiotherapy or immunotherapy in clinical practice to improve tumor treatment in the near future. In addition, similar to genome editing, external transcriptome editing can restore or remove functional essential m6A sites that are mutated or maladjusted in cancer, and such editing may also have clinical applications in future cancer treatments. In conclusion, studying the modification of m6A in cancer is a new frontier in cancer research. It not only reveals new aspects of epigenetic regulation in cancer, thereby contributing to the understanding of tumogenesis, immune response, and drug resistance, but will also lead to the development of effective new therapies. Targeting malregulated m6A regulators with effective inhibitors (or targeting mutated or dysfunctional m6A sites by targeting external transcriptomic editing), alone or in combination with other therapies, may have potential therapeutic potential for treating various types of cancer, particularly those that are resistant to existing therapies.
RNA m6A modification has attracted increasing attention as a prime focus for epigenetic research, and its involvement in various biological processes and disease progression is also being increasingly investigated. From an epigenetic perspective, m6A modification provides new insights into the pathogenesis of many diseases, especially tumours. m6A plays a dual role by either promoting or inhibiting the occurrence and development of tumours by regulating mRNA and ncRNA levels of oncogenes or tumour suppressor genes [102]. This article reviews the role, mechanism and clinical applications of m6A in esophageal cancer. Recent studies have demonstrated that the role of m6A modification in the occurrence, development and biological function of esophageal cancer is complex, which is a novel research direction in epigenetics. The complexity of m6A modification may be related to different tumour stages, tissue types and complex RNA metabolism in the inflammatory and immune microenvironments of esophageal cancer. Furthermore, m6A modification is also a key factor in the diagnosis, treatment and prognosis of patients with esophageal cancer. Several studies have been conducted to investigate the effect of m6A modification on esophageal cancer, and some m6A factors have been reported to regulate immune filtration and serve as diagnostic and prognostic markers and therapeutic targets for esophageal cancer; however, certain concerns regarding their function warrant further investigation. m6A modification may serve as an early diagnostic marker and a therapeutic target for esophageal cancer. METTL3, hnRNPA2B1, ALKBH5 and FTO have been confirmed to be upregulated in esophageal cancer and are expected to act as biomarkers for the diagnosis and prognosis of esophageal cancer. However, current literature is still preliminary and limited, and sufficient samples have not been obtained for testing. Therefore, further studies and large-scale clinical trials of m6A are required. Furthermore, mutations in some m6A regulatory factors increase the complexity of m6A-mediated epigenetic modification regulation. For instance, as a YTHDC2 risk variant, rs2416282 alters YTHDC2 expression and reduces ESCC risk [76]. Currently, no other studies have further investigated the mechanism of m6A modification variation, and its complex mechanism in esophageal cancer remains unknown. In addition, current research on m6A modification is limited to the interaction mechanism of esophageal cancer, and studies on medical transformation remain elusive. Moreover, targeted therapy and immunotherapy offer great prospects for the treatment of esophageal cancer [103]. Therefore, more translational studies are required to promote the clinical applications of m6A modification, such as the combination of m6A with chemotherapy and immunotherapy. In conclusion, research on m6A modification in esophageal cancer is progressing rapidly and is a prospective approach for investigating clinical medicine transformation in future studies. Finally, the immune system is the host's defense against infection and disease. Meanwhile, immunotherapy is a new cancer treatment strategy, which has been widely used to treat various solid tumors, including various gastrointestinal tumors. In recent years, m6A regulatory factors have been widely studied in tumor immunotherapy and immune evasion. In addition, tumor immunotherapy is the most promising therapeutic strategy. Currently, immune invasion mediated by m6A modification is becoming a hot field of studying the pathogenesis and prognosis of esophageal cancer. However, the study of immune invasion mediated by m6A modification in esophageal cancer is still in its infancy, and m6A modification is expected to make new breakthroughs in this field in future studies.
All authors listed have significantly contributed to the development and the writing of this article.
This work was supported by the grants of Taizhou Central Hospital (Taizhou University Hospital) No. 2019KT047.
No data was used for the research described in the article.
The authors declare no conflict of interest.
No additional information is available for this paper. |
|
PMC9647494 | Jaayke L. Fiege,Benedikt Hirt,Volker Gräf,Stefan Nöbel,Dierk Martin,Jan Fritsche,Katrin Schrader,Mario Stahl | Uridine as a non-toxic actinometer for UV-C treatment: influence of temperature and concentration | 07-11-2022 | UV-C,Photochemical actinometry,Uridine,Quantum yields,UV-C reactor,Iodide/iodate actinometry | UV-C treatment is an effective method to inactivate microorganisms and therefore gets increasingly more attention in food industry, especially for liquid products. To test and monitor different UV-C reactor designs, a photochemical actinometer is required that gives reliable UV-C dose values and is non-toxic allowing frequent control of the production chain. Here, a variable concentrated aqueous uridine solution is tested as a photochemical actinometer. Uridine reacts at 262 nm by photohydration to a single photoproduct not absorbing any light. A concentration dependent quantum yield (Ф) was quantified in the range of 0.2–3.0 mM uridine. Results show that uridine is as accurate as the commonly accepted iodide/iodate actinometry, but not as precise. Especially at higher concentrations a higher number of measurements becomes necessary. Further, a temperature correction is presented for 10 °C > ϑ > 30 °C. Taking these results into account, uridine can certainly be considered as a non-toxic dosimeter for UV-C systems. | Uridine as a non-toxic actinometer for UV-C treatment: influence of temperature and concentration
UV-C treatment is an effective method to inactivate microorganisms and therefore gets increasingly more attention in food industry, especially for liquid products. To test and monitor different UV-C reactor designs, a photochemical actinometer is required that gives reliable UV-C dose values and is non-toxic allowing frequent control of the production chain. Here, a variable concentrated aqueous uridine solution is tested as a photochemical actinometer. Uridine reacts at 262 nm by photohydration to a single photoproduct not absorbing any light. A concentration dependent quantum yield (Ф) was quantified in the range of 0.2–3.0 mM uridine. Results show that uridine is as accurate as the commonly accepted iodide/iodate actinometry, but not as precise. Especially at higher concentrations a higher number of measurements becomes necessary. Further, a temperature correction is presented for 10 °C > ϑ > 30 °C. Taking these results into account, uridine can certainly be considered as a non-toxic dosimeter for UV-C systems.
UV-C treatment is of interest in food industry mainly to replace or complement classical pasteurization. Instead of high temperatures, the microorganisms are devitalized by changes in their DNA that cannot be repaired. UV-C energy (approx. 254 nm) leads to the formation of thymine cyclobutene dimers and thus to sterile germs (Beukers and Berends, 1960; Setlow and Setlow, 1962; Swenson and Setlow, 1963). The distinct CO2 footprint by thermal treatment methods is a strong argument for using low energy consuming UV-C technique in today's food industry (Tran and Farid, 2004; Pennell et al., 2008; Wang et al., 2010; Bandla et al., 2012; Baysal et al., 2013; Flores-Cervantes et al., 2013; Gayán et al., 2013; Groenewald et al., 2013; Cilliers et al., 2014; Ansari et al., 2019). But not just the germicidal effect by UV-C is sensed by the food industry. Current research also focuses on the improvement of certain foods via UV treatment such as vitamin D3 formation in milk out of its constituents (EFSA, 2016). To avoid partly excessive and/or insufficient treatment of liquid food with UV energy a homogeneous energy distribution is required. Flow-through reactors either need to mix the medium homogeneously using, e.g., turbulent flows, or they need to use laminar flows that are thin enough to allow sufficient penetration of UV-C energy. Especially for turbid or opaque liquids, the latter treatment is challenging. In order to measure the energy input and therefore dose (J/L) that liquid food would experience during treatment in different reactors, an accurate dose measurement is necessary. This is usually done by applying a photometric actinometer. In general, a chemical actinometer utilizes a specific photochemical reaction to determine incident energy dose into a defined volume, which requires exactly known quantum yields (Ф). Quantum yield (Ф) is the number of affected molecules divided by the number of absorbed photons at a specific wavelength (Kuhn et al., 2004; Rabani et al., 2021). The application of chemical actinometric solutions allows the determination of the exact energy release into complex geometries (Kuhn et al., 1989) such as liquids flowing through UV reactors. There are several chemical actinometers published working in the UV-C range (Kuhn et al., 2004; Rabani et al., 2021). However, most actinometers include toxic chemicals or are complicated to use which may work in laboratories but remains challenging for food-producing companies. An easy-to-use and food grade actinometer is required to regularly monitor energy dose values in UV-C food treatment reactors. Therefore, this study focuses on aqueous uridine solution as a chemical actinometer. Uridine is non-toxic as it is one of the unmodified nucleosides usually found in RNA. Uridine contains the chromophore uracil that has an absorption maximum at approx. 262 nm which, when in aqueous solution, depletes upon UV-C (254 nm) irradiation. This degradation follows (pseudo-) first order kinetics (Wang 1962; von Sonntag and Schuchmann, 1992; Cataldo 2017). The photohydration was identified at the 5–6 double bond at the carbonyl group of the photoproduct (Moore, 1958) that does not absorb any light, neither at 262 nm nor any other wavelength in the UV/Vis range (Figure S1; Moore and Thomson, 1955). This bleaching is even sixteen times more pronounced for the nucleoside uridine when compared to the pure pyrimidine base uracil (Sinsheimer and Hastings, 1949). At neutral pH and 20 °C the photoproduct 6-hydroxy-5,6-dihydrouridine, also called uridine hydrate, is stable for about 150 h (Fisher and Johns, 1976). This, the general availability, low costs (e.g., 200 €/100 g at Carl Roth GmbH, 2022), simple handling (no acids and bases are required), and the non-toxicity of uridine makes it a perfect candidate for a chemical actinometer in food industry using UV-reactor systems. However, the reliability and accuracy of a chemical actinometer are highly dependent on the applied quantum yields. To date, quantum yields are only published for very low uridine concentrations (approx. 10−4 Mol/L). At an incident wavelength of 254 nm quantum yields for the photohydration of uridine range between 0.017 and 0.022 Mol/einst (Sinsheimer, 1954; Swenson and Setlow, 1963; Görner, 1991; von Sonntag and Schuchmann, 1992; Gurzadyan and Görner, 1996; Linden and Darby, 1997; Zhang et al., 1997; Jin et al., 2006). While it is generally accepted that the quantum yield of uridine is relatively independent of the incident wavelength, as long as it is in the range of 238–280 nm (Swenson and Setlow, 1963; Rahn and Sellin 1979), the only important factor for uridine degradation relies on the probability whether a photon is or is not absorbed for photohydration (Setlow and Setlow, 1961; Jin et al., 2006). This probability decreases with decreasing incident intensities, i.e. <1 mW/cm2 (Linden and Darby, 1997). Additionally, smaller uridine concentrations are said to give more accurate dose values due to Taylor series expansion of the Lambert-Beer law used for calculation of dose values (Jin et al., 2006). However, low concentrated uridine solutions have low absorbances and allow deep UV-C penetration depths into the solution. When penetration depth of the actinometric solution overcomes the width of the treatment space, energy would get lost instead of being measured by the actinometric solution. In order to also measure thin laminar flows a higher uridine concentration becomes necessary when penetration depth is larger than width of the thin film. An actinometric solution with higher uridine concentration would also allow examining large UV doses without falling below the photometric detection limit. However, quantum yields of higher uridine concentration (>10−4 Mol/L) were to date not investigated to the best of our knowledge. Thus, quantum yields dependent on higher uridine concentrations (10−4 Mol/L< cUridin < 10−2 Mol/L) were investigated in this study as well as compared with standard methods such as iodide/iodate actinometry in different UV-C reactor designs including also a laminar thin film reactor with fluid guiding elements (FGE) generating thin films of 0.06 cm (Gök et al., 2021; Hirt et al., 2022a, Hirt et al., 2022b).
Uridine solutions with variable concentrations were prepared by mixing different masses of uridine (Carl Roth GmbH, Karlsruhe, Germany) into demineralized water. The aimed concentrations were 50, 100, 200, 350, 500 and 750 mg/L, i.e. 0.2, 0.4, 0.8, 1.4, 2.0 and 3.0 mM. In-house built UV-C reactors (Max Rubner-Institut, Karlsruhe, Germany) were used for quantification and validation of the uridine actinometry. All UV-C reactors contain low-pressure mercury lamps with an emission peak at approx. 254 nm. Flow rates for all experiments were adjusted by using a peristaltic pump (Pumpdrive S206; Heidolph Instruments GmbH & Co. KG, Schwabach, Germany) and the uridine solution was pumped five times completely through each reactor. The sample solutions were tempered during the experiments at 20 ± 2 °C (or any other desired temperature) using a coil in a cryostat bath (F32; Julabo GmbH, Seelbach, Germany). Samples were taken after the preparation of the solution, immediately before UV-C treatment, and after each pass through the reactor. Absorbances after preparation and before UV-C treatment, respectively, remained unchanged proving the stability of the solution to visible light. To prevent mixing with residuals that remained in the reactor from the prior pass, 80 or 200 ml of each pass were discarded before sampling from the coiled tube or straight tube and thin film reactor, respectively. Each uridine concentration was tested in triplicate and each experiment was pumped through the system in five passes generating five data points for each initial concentration. The spread of each data point is given by the standard deviation of n = 3. For comparison exactly the same procedure was conducted for each UV-C reactor type using iodide/iodate actinometry originally developed by Rahn (1997). The iodide (0.6 M) – iodate (0.1 M) solution in 0.01 M borate buffer absorbs all radiation below a wavelength of 290 nm and generates triiodide as a result of a photochemical reaction with UV-C photons (Rahn et al., 2003). Triiodide exhibits an absorbance maximum at 352 nm and its linear formation kinetic allows an easy quantification of absorbed UV-C dose (D in J/L). This actinometric technique is accepted as a standard actinometer (Rabani et al., 2021) and the here reported results are achieved by following the protocols described elsewhere (e.g., Müller et al., 2014, Gök et al., 2021, Hirt et al., 2022a,b).
This reactor contains a 40 W low-pressure mercury lamp with a specified output of 15 W UV-C (TUV-36-T5; Philips GmbH, Hamburg, Germany) that is surrounded by a 23 m long UV-transparent plastic tube (fluorethylenpropylen: FEP) coiled over a length of 750 mm. The inner tube diameter (di) is 3.7 mm and the inner diameter of the coil is 38.5 mm. Due to centrifugal forces, the medium suppresses liquid from the outer wall so that continuous Dean vortices are generated, when Reynolds numbers (Re) are smaller than the critical Reynolds number (Recrit) and above Dean number values (De) of Decrit = 54 (Dean, 1928; Hämmerlin, 1957). This system has a higher critical Reynolds number (Recrit) than simple tube flows with Recrit of approx. 2,300. Recrit in a coiled tube can be calculated using Eq. (1) (Gnielinski, 2010).Recrit of the coiled tube reactor in this study is 9194. The Reynolds number of an aqueous solution in the coiled tube reactor at 30 L/h is calculated using Eq. (2) and accounts for 2868, which is significantly below Recrit of the coiled tube reactor (9,194).υ is the velocity (m/s), d is the diameter of the tube (0.0037 m), and ρ is the density (approx. 1,000 kg/m3) and η the dynamic viscosity of the aqueous solution (approx. 1.0 mPa s) at 20 °C. Dean number (De) can be calculated from Re and the inner diameters of the tube (di) and the one of the coil (D) (Eq. (3)). The respective Dean number for an aqueous solution at 30 L/h is De = 889, which clearly exceeds Decrit = 54. Thus, all experiments at a volume flow of 30 L/h are certainly within the regime of Dean vortices.
The straight tube reactor contains 24 straight tubes having an inner diameter of 6 mm and an outer diameter of 6.6 mm. These tubes consist of UV-C transparent fluorethylenpropylen (FEP) and are connected with non-UV-transparent U-turns. The total length is 19.3 m of which 16.1 m can be penetrated by the UV light. This tube system is arranged in a commercial UV box (BS04 UV box; UV Messtechnik Opsytec Dr. Gröbel GmbH, Ettlingen, Germany). Twenty 18 W low-pressure mercury lamps with an specified UV-C output of 4.5 W each are positioned 300 mm above the FEP tubes. Thus, the efficiency of the UV lamps is not influenced by the temperature of the medium allowing the investigation of the temperature dependence of the quantum yield. The straight tube reactor can be run at laminar (Re < 2,300) and turbulent flow conditions (Re > 2,300) by adjusting the flow rate of the medium. Re is calculated in the straight tube systems according to Eq. (2) with diameter d of the straight tube (0.006 m). Hence, when volume flow exceeds 40 L/h, the regime changes from laminar to turbulent flow. Also, the intensity of the UV-C energy of the lamps can be electronically dimmed from 100 to 1 % allowing the investigation of different UV-C intensities independent of the flow rate. The punctual irradiance (mW/cm2) can be measured with the radiometric sensor included in the commercial UV box (BS04 UV box; UV Messtechnik Opsytec Dr. Gröbel GmbH, Ettlingen, Germany).
The core of the thin film reactor (TFR) is a 20 W low-pressure mercury lamp (UVpro N20-2, orca GmbH) with a 7.5 W UV-C output. The medium is pumped with 60 L/h as a laminar thin film of ca. 3.1 mm between the quartz glass cylinder of the lamp and the stainless-steel sleeve over a length of 34.6 cm. To improve the efficiency of UV-C irradiation into opaque liquid foods (e.g. milk) fluid guiding elements (FGE) are inserted into the annular gap (Gök et al., 2021, Hirt et al., 2022a, Hirt et al., 2022b). The FGE are made of stainless steel and were designed by the Institute of Micro Process Engineering of the Karlsruhe Institute of Technology (KIT) and manufactured using selective laser melting. These FGE divide the original liquid flow into three partial flows with smaller diameters that are guided alternately towards the UV-C source. Hansjosten et al. (2018) already showed the efficiency of FGE used in pipe-in-pipe heat exchanger wherein the length of the heat exchanger was able to be reduced by ten times. When using FGE in this study the resulting gap between quartz glass and inner wall of the FGE is 0.6 mm, which significantly increases the UV-C efficiency in laminar flowing liquids with high absorbances shown by Gök et al. (2021) and Hirt et al., 2022a, Hirt et al., 2022b.
All samples were measured in disposable semi-micro UV cuvettes with an optical path length of 10 mm using a Unicam UV2-100 UV/Vis Spectrometer immediately after irradiation of the samples (<20 min). The absorbance was measured at the maximum of the uridine peak (approx. 262 nm). To achieve reliable results, samples were diluted concerning their initial concentrations to keep maximum absorbances <1.5. Each sample was measured in triplicate (n = 3).
To calculate UV doses, the extinction coefficient of the uridine solution is necessary. Since the photoproduct of uridine, uridine hydrate, does not absorb any light the spectrometric detection at solely 262 nm, is sufficient to quantify the photohydration (e.g., Rabani et al., 2021). Photometric scans from 230 to 900 nm of a highly concentrated uridine solution (750 mg/L = 3 × 10−3 Mol/L) before and after UV-C treatment prove that no other light absorbing products were formed (Figure S1). Literature data on extinction coefficients of uridine vary between 8,000 and 10,185 L mol−1 ·cm−1 (von Sonntag and Schuchmann, 1992; Jin et al., 2006; Cataldo 2017). Here, 16 different uridine concentrations ranging between 0.0041 and 500 mmol/L, i.e. 1 and 125,000 mg/L, were measured to determine the extinction coefficient. In order to keep absorbances below 1.5 each concentration was diluted accordingly and measured in triplicate. The three linear regression slopes gave an average extinction coefficient of 9,560.7 L mol−1 · cm−1 with a standard deviation of 50.9 L mol−1 · cm−1 (Figure S2) that fits well within the data published previously (von Sonntag and Schuchmann, 1992; Jin et al., 2006; Cataldo 2017).
This study aimed to develop a chemical actinometer that can be applied independent of the knowledge about the geometry of a reactor and, thus, the total fluence or dosage per 1 L of actinometric solution passing through the reactor ([D] = J·L−1) is investigated here instead of spatial and time-resolved irradiance or fluence rate ([E] = J·cm−2·s−1). To do so, the equation given by Zhang et al. (1997) was modified similar to the equation given by Rahn (1997) for the potassium iodide actinometer. However, the uridine actinometer follows a reaction of pseudo first-order kinetics (Figure S3 and Wang, 1962; von Sonntag and Schuchmann, 1992; Cataldo, 2017), while the potassium iodide actinometer linearly depends on the formation of triiodide. Thus, the numerator of Eq. (4) needs to be logarithmized in contrast to the potassium iodide equation by Rahn (1997).A0 and Ai are the dimensionless absorbances at 262 nm measured before and after treatment through the reactor, respectively, pl is the pathlength of the cuvette (1 cm), Ф is the quantum yield (mol/einst), ελ is the molar extinction coefficient (L mol−1 cm−1) at irradiation wavelength λ, i.e. here 254 nm. The molar extinction coefficient at 254 nm is approximately 85 % of the molar extinction coefficient measured for 262 nm (Figure S1). To achieve the absolute dosage D (J/L) per pass through the reactor the photon flux Pλ (J/einst) at the irradiation wavelength λ is also required (Eq. (5)):h is the Planck constant (6.626 × 10−34 m2 kg s−1), c is the speed of light (3 × 108 m s−1) and λ the respective wavelength (254 × 10−9 m). When incident wavelength is 254 nm, Pλ becomes 4.716 × 105 J/einst. Multiplication of Pλ with Eq. (4) results in the absolute dosage D (J/L):
The absorbance Ai changes with each pass through the reactor, while the delivered energy from the UV-C source is constant. Since the remaining parameters are also constants, quantum yield (Ф) is the only parameter that is allowed to change in Eq. (6) by altering concentration. Thus, quantum yield is dependent on uridine concentration. The dose was previously measured using an often applied, but toxic and, therefore, unfeasible chemical actinometer for the food industry - the iodide/iodate actinometry according to Rahn (1997). The iodide/iodate actinometry was proven as a reliable reference that can be used to determine quantum yields of other chemical actinometers (Goldstein and Rabani, 2008; Rabani et al., 2021). Measuring the dose with iodide/iodate at exactly the same conditions (30 L/h, 20 °C) with triplicates à five passes resulted in 856 ± 54 J/L (n = 3) (Figure S4). Using this value as dose D for the coiled tube reactor, Ф for each pass of all experiments was able to be calculated by converting Eq. (6) in terms of Ф. The resulting Ф were plotted versus the concentration that was measured from the absorbance. It is noteworthy that after each pass of each experiment the concentration depletes slightly, which is why each initial concentration resulted in five separate datapoints (Figure 1). That is, six different initial uridine concentrations resulted in 30 datapoints that can be empirically modeled by a decreasing power function (Eq. 8) with an R2 of 0.82. Using Eq. (7) a quantum yield can be calculated for any concentration (c in Mol/L) between 50 and 750 mg/L urdidine. The average of all initial uridine concentrations in the coiled tube reactor resulted in 986 J/L with a standard deviation of ±82 J/L (Figure 2), where each initial concentration (n = 6) was measured as a triplicate of five passes. This dose is higher than measured with iodide/iodate actinometry (856 ± 54 J/L; triplicate (n = 3) à five passes) but the discrepance is not significant (p = 0.30) according to ANOVA test. However, the already small quantum yield of uridine decreases even more with increasing concentration when compared to iodide/iodate actinometry with a relatively constant and significantly higher Ф of approximatly 0.73 mol/einst (Rahn, 1997). Thus, propability for uridine hydration may deplete with increasing concentration also affecting the precision of uridine actinomatery at higer concentrationsas it was already observed in the literature (e.g. Jin et al., 2006). An additional reason for the unprecise results gained by high uridine concentration may be the sharper absorption bands in spectrometric measurements, when compared to the rather broader peaks by lower uridine concentrations (Figure S1). Sharper peaks may have more significant deviations due the unpreventable polychromacity when using a light dispersing element with a slit (monochromator). Lower concentrations and thus broader peaks are less susceptible to these random errors caused by minor misadjustments of the monochromator and cause more stable or precise results. However, accuracy seems to remain at higher concentrations. The difference of accuracy and precision is graphically illustrated in the supplementary (Figure S7). That is, the result of many measurements averages in statistically the same value 986 ± 82 J/L (Figure 2) (p = 0.97). Thus, as long as a sufficient number of measurements is available, high uridine concentrations (≤750 mg/L) are feasible to be applied as a relatively robust and safe actinometer, when toxic actinomteres need to be avoided.
In a straight tube reactor, where medium was treated completely turbulent (volume flow: 100 L/h), i.e. Reynolds number >2300, the absorbances of the different media (iodide/iodate: A254nm = 180 and uridine solution: 30 > A254nm > 2) should not influence the dose calculation at all. Here, the power of the lamp was varied at 30, 50, 70, and 100%. An initial concentration of 100 mg/L uridine was chosen for different intensities. The dose values using iodide/iodate actinometry after Rahn (1997) (Figure S5) and the ones using uridine actinometry suggested in this work (Eqs. (6) and (7)) were plotted in Figure 3 against the irradiance (mW/cm2) measured with the stationary radiometric sensor. Iodide/iodate actinometry and uridine actinometry have an overlap with a slope discrepancy of less than 5%. Five measurements à five passes were also conducted with a higher concentration of 500 mg/L uridine at 100% lamp power showing a decreasing certainty, i.e. decreasing precision of uridine actinometry with increasing uridine concentration as shown in Section 3.3. However, accuracy remains, i.e. five measurements à five passes with high uridine concentration (500 mg/L) average in the same dose value as one measurement à five passes with lower uridine concentration (100 mg/L).
When working with reactors using laminar flowing thin films, pentration depth of the energy into the actinometric solution becomes of high importance, which is in contrast to the previous mentioned reactor desgins. The penetration depth (dp) is the distance at which the intensity of the incoming energy is reduced down to 1/e = 37 % of the initial energy, as the intensity of an electromagnetic wave decays exponentially inside a medium (Lambert-Beer law). dp can be calculated from absorbance (A) at according wavelength measured with a photometer (Eq. 8): The FGE equipped reactor used in this study has a significant smaller treatment width (0.6 mm) than dp of 254 nm into a 50 mg/L Uridine solution (dp = 2.2 mm) but a larger width than penetration depth into iodide/iodate solution (dp = 0.02 mm). The exponential energy decay for a radial energy source into different uridine concentrations can be calculated using Lambert-Beer law according to Koutchma and Arisi (2004) and Hirt et al., 2022a, Hirt et al., 2022b (Eq. (9)): The results are illustrated in Figure 4 assuming I0 as 100 % at the outer quartz glass sleeve surrounding the low-pressure mercury lamp (r0 = 1.159 cm) with a maximum penetration depth of 0.063 cm (rmax = 1.222 cm) (Hirt et al., 2022a, Hirt et al., 2022b). All energy beyond a penetration depth of 0.063 cm, i.e. rmax = 1.222 cm, is lost to the wall of the FGE and cannot be measured by the actinometric uridine solution. Thus, an “underconcentrated” uridine solution would lead to underestimation of the doses in such laminar thin film system. In order to use uridine solution as reliable chemical actinometer to study/control laminar thin film systems, the uridine concentration needs to be examined at which calculated doses agree with the one of iodide/iodate actinometry. The doses at a volume flow adjusted to 60 L/h were measured with different uridine concentrations as well as with iodide/iodate actinometry resulting in 329 ± 3 J/L for the latter (Figure S6). Each uridine concentration was measured in triplicates à five passes. The results are displayed in Figure 5 together with the theoretical measurable doses from Eq. (9) with r0 = 1.159 cm and r = 1.222 cm. It is obvious that actinometric solutions with low uridine concentrations (here <500 mg/L) would significantly underestimate the actual doses applied to the passing medium in this reactor. Thus, higher uridine concentrations are necessary to measure actinometric dose more accurately in laminar thin film systems, although precision depletes with higher concentrations. The required uridine concentration to achieve reliable results depends on the width of the treatment space. Here, a concentration of ≥500 mg/L is necessary to achieve similar results as with iodide/iodate actinometry.
The effect of the temperature of the medium on the quantum yield was invastigated on the straight tube reactor, where the UV-C lamps are relatively far away from the treated medium (300 mm) and thus are not influenced by the temperature of the treated medium. Uridine actinometry was conducted in triplicates à five passes with 500 mg/L and a turbulent volume flow of 100 L/h. Temperatures were adjusted to three different values (approx. 10, 20, and 30 °C). These temperatures are plotted against the resulting doses from uridine actinometry using Eqs. (6) and (7) in Figure 6. In addition, the dose from iodide/iodate actinometry (342 ± 7 J/L; n = 3) was measured under the same conditions. Since, the calculation by iodide/iodate actinometry already includes a temperature correction (Rahn, 1997), the iodide/iodate data measured at 19.3 ± 0.3 °C is plotted as a constant. A positive linear relationship to temperature becomes visible for uridine actinometry. Thus, the measured dose using Eqs. (6) and (7) can be corrected for the applied temperature using the empirical Eq. (10): Dmeas is the measured dose as calculated from Eqs. (6) and (7) in J/L, ϑ is the experimental temperature in °C, and Dcorr is the temperature corrected dose in J/L. This correction stays valid in the range of 10–30 °C and for UV-C reactors where lamp power remains unaffected of the temperature of the treated medium. When UV-C lamps are more proximate to the treated medium, an additional influence of temperature on efficiency of UV-C lamp may arise.
If a food-grade actinometric system is required uridine actinometry can be applied as an alternative to the toxic iodide/iodate actinometric system. The results of this study show that concentration of uridine can be varied from 0.2 to 3.0 mM using the here presented concentration dependency for the calculation of the quantum yield . This novel application with also higher concentration (i) extends the total detectable UV-C dosage range and (ii) enables measurements in laminar thin film reactor designs, where width of treatment space is <2.2 mm, which is the penetration depth (dp) of 254 nm into a commonly concentrated actinometric uridine solution (approx. 0.2 mM uridine). However, as quantum yield decreases with increasing uridine concentration the precision of uridine actinometetry also depletes. Thus, concentration of uridine should be kept as low as possible in order to receive more precise results with low standard deviations. But when concentration needs to be increased e.g. due to the above mentioned reasons, one could simply increase the number of measurements, which is cheap and easy due to low costs and non-toxic behavior of uridine. The results may be not as precise as measured with low concentrated uridine solutions or with iodide/iodate actinometry, but the accuracy remains, i.e. th average of many measurements shall be statistically the same. Additionally, experimental temperature seems to influence the measured dose by uridine actinometry, but measured dose results can be corrected accordingly (10–30 °C): .
Jaayke L. Fiege: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Wrote the paper. Benedikt Hirt; Volker Gräf; Dierk Martin; Katrin Schrader: Conceived and designed the experiments; Analyzed and interpreted the data; Wrote the paper. Stefan Nöbel: Conceived and designed the experiments; Wrote the paper. Jan Fritsche: Analyzed and interpreted the data; Wrote the paper. Mario Stahl: Contributed reagents, materials, analysis tools or data; Wrote the paper.
Mario Stahl was supported by Forschungskreis der Ernährungsindustrie [AiF 21130N].
Data included in article/supp. material/referenced in article.
The authors declare no conflict of interest.
No additional information is available for this paper. |
|
PMC9647495 | 36321792 | Pier Paolo Leoncini,Patrizia Vitullo,Sofia Reddel,Valeria Tocco,Valeria Paganelli,Francesca Stocchi,Elena Mariggiò,Michele Massa,Giovanni Nigita,Dario Veneziano,Paolo Fadda,Mario Scarpa,Martina Pigazzi,Alice Bertaina,Rossella Rota,Daria Pagliara,Pietro Merli | MicroRNA profiling of paediatric AML with FLT-ITD or MLL-rearrangements: Expression signatures and in vitro modulation of miR-221-3p and miR-222-3p with BRD4/HATs inhibitors | 31-10-2022 | acute myelogenous leukaemia,microRNAs,epigenetic drugs,biomarkers,children | Novel therapeutic strategies are needed for paediatric patients affected by Acute Myeloid Leukaemia (AML), particularly for those at high-risk for relapse. MicroRNAs (miRs) have been extensively studied as biomarkers in cancer and haematological disorders, and their expression has been correlated to the presence of recurrent molecular abnormalities, expression of oncogenes, as well as to prognosis/clinical outcome. In the present study, expression signatures of different miRs related both to presence of myeloid/lymphoid or mixed-lineage leukaemia 1 and Fms like tyrosine kinase 3 internal tandem duplications rearrangements and to the clinical outcome of paediatric patients with AML were identified. Notably, miR-221-3p and miR-222-3p resulted as a possible relapse-risk related miR. Thus, miR-221-3p and miR-222-3p expression modulation was investigated by using a Bromodomain-containing protein 4 (BRD4) inhibitor (JQ1) and a natural compound that acts as histone acetyl transferase inhibitor (curcumin), alone or in association, in order to decrease acetylation of histone tails and potentiate the effect of BRD4 inhibition. JQ1 modulates miR-221-3p and miR-222-3p expression in AML with a synergic effect when associated with curcumin. Moreover, changes were observed in the expression of CDKN1B, a known target of miR-221-3p and miR-222-3p, increase in apoptosis and downregulation of miR-221-3p and miR-222-3p expression in CD34+ AML primary cells. Altogether, these findings suggested that several miRs expression signatures at diagnosis may be used for risk stratification and as relapse prediction biomarkers in paediatric AML outlining that epigenetic drugs, could represent a novel therapeutic strategy for high-risk paediatric patients with AML. For these epigenetic drugs, additional research for enhancing activity, bioavailability and safety is needed. | MicroRNA profiling of paediatric AML with FLT-ITD or MLL-rearrangements: Expression signatures and in vitro modulation of miR-221-3p and miR-222-3p with BRD4/HATs inhibitors
Novel therapeutic strategies are needed for paediatric patients affected by Acute Myeloid Leukaemia (AML), particularly for those at high-risk for relapse. MicroRNAs (miRs) have been extensively studied as biomarkers in cancer and haematological disorders, and their expression has been correlated to the presence of recurrent molecular abnormalities, expression of oncogenes, as well as to prognosis/clinical outcome. In the present study, expression signatures of different miRs related both to presence of myeloid/lymphoid or mixed-lineage leukaemia 1 and Fms like tyrosine kinase 3 internal tandem duplications rearrangements and to the clinical outcome of paediatric patients with AML were identified. Notably, miR-221-3p and miR-222-3p resulted as a possible relapse-risk related miR. Thus, miR-221-3p and miR-222-3p expression modulation was investigated by using a Bromodomain-containing protein 4 (BRD4) inhibitor (JQ1) and a natural compound that acts as histone acetyl transferase inhibitor (curcumin), alone or in association, in order to decrease acetylation of histone tails and potentiate the effect of BRD4 inhibition. JQ1 modulates miR-221-3p and miR-222-3p expression in AML with a synergic effect when associated with curcumin. Moreover, changes were observed in the expression of CDKN1B, a known target of miR-221-3p and miR-222-3p, increase in apoptosis and downregulation of miR-221-3p and miR-222-3p expression in CD34+ AML primary cells. Altogether, these findings suggested that several miRs expression signatures at diagnosis may be used for risk stratification and as relapse prediction biomarkers in paediatric AML outlining that epigenetic drugs, could represent a novel therapeutic strategy for high-risk paediatric patients with AML. For these epigenetic drugs, additional research for enhancing activity, bioavailability and safety is needed.
Acute Myeloid Leukaemia (AML) constitutes 20% of all paediatric leukaemia and is responsible for substantial mortality. Despite progress made over the past years in diagnosis, risk stratification and treatment of AML, survival remains suboptimal with a success rate of 60–70% (1–3), with relapse being the leading cause of death. Risk stratification in patients with AML is of paramount importance in order to deliver tailored therapy, enabling treatment intensification in high-risk patients (for example, allogenic stem cell transplantation in high-risk in first complete remission). Moreover, novel therapies are needed in order to further improve prognosis in these patients. Among prognostic factors, cytogenetic and molecular abnormalities, together with Minimal Residual Disease and treatment response, play a pivotal role in defining AML prognosis and treatment (4,5). Rearrangements involving Histone-lysine N-methyltransferase 2A (KMT2A) gene, formerly known as myeloid/lymphoid or mixed-lineage leukaemia 1 (MLL1) gene, as well as Fms like tyrosine kinase 3 (FLT3) internal tandem duplications (FLT3-ITD) represent useful prognostic factors and, possibly, therapeutic targets. Indeed, they still define an intermediate to high risk AML (1–3,6). In particular, FLT3-ITD mutations are associated with higher risk of relapse and dismal prognosis (1,3,7), whereas MLL-rearranged AML is an heterogeneous group of diseases with more than 100 rearrangements being described and with different outcome largely dependent on the fusion partner (8). FLT3-ITD and MLL rearrangements have been functionally linked to dysregulation of expression of microRNAs (miRNAs or miRs) (9). miRs are small non-coding RNA molecules (~18-22 nucleotides long), involved in several cellular processes (10,11). In cancer, they are implicated both in promoting carcinogenesis (oncomiRs) and in suppressing tumour transformation (12). Their role in AML have been extensively investigated over the past years reviling promising data on diagnosis, prognostic stratification and, possibly, treatment in AML patients (13–15). Despite extensive research performed to understand the role of miRs in AML, the majority of studies are focused on adult patients, while a precise characterization of miRNAs expression in paediatric AML is less documented. Moreover, data on the role of different miRs are conflicting because of the variability of genetic abnormalities found in AML (16). In the present study, it was aimed to identify AML specific miR signatures in a cohort of patients harbouring molecular lesions (FLT3-ITD and MLL rearrangement), studying the expression of distinct miR sets in relation to relapse risk. Epigenetic networks, including histone modification mechanisms, are involved in the regulation of both miRs expression and function. An increasing interest in the field of cancer therapeutic drugs is focused on small molecular compounds targeting epigenetic regulation (17). Bromodomain and extra-terminal domain family of proteins (BET) and histone acetyl transferase (HAT) inhibitors proteins are the best characterized ones. BET are effective in preventing Bromodomain Containing protein 4 (BRD4) associated transcription of several oncogenes, reducing proliferation and increasing apoptosis in AML (18–22). BRD4 is a member of BET family proteins, characterized by the presence of functional structures called bromodomains which bind specific acetylated residues on histone tails to modulate transcription of target gene (23–25), enhancing transcription of several oncogenes (26). Among BRD4 inhibitors, JQ1 was used as its activity on modulation of miRs was previously described (27). JQ1 [(S)-tert-butyl-2-(4-(4-chlorophenyl)-2,3,9-trimethyl-6H-thieno[3,2-f][1,2,4]triazolo[4,3-a][1,4]diazepin-6-yl)acetate] is a small molecule belonging to the thienotriazolodiazepine group and it prevents the binding of BRD4 to acetylated residues on histone H3 tails, particularly H3AcK14 (20,28–30). Furthermore, among HAT inhibitors, curcumin, a natural compound extracted from the root of Curcuma Longa has been shown to inhibit acetylation of histone tails, blocking the activity of the HAT p300 even causing its proteasomal degradation (31). This results in a global decrease of acetylation on histone tails and a consequent modulation of gene transcription (32,33). BRD4 inhibitors have exhibited only moderate results in clinics and novel ways to increase their antitumour activity are needed (34). It was therefore hypothesized that the association with curcumin would increase JQ1 efficacy. The BET family are ‘readers’ of chromatin acetylation whereas HAT could be classified as a ‘writer’ of histone acetylation (34) thus offering a theoretical basis for JQ1 and curcumin synergic activity. Moreover, it was previously showed that also JQ1, like curcumin, blocks p300-mediated acetylation (25,35). Thus, it was investigated in vitro whether a combination of BRD4 and HAT inhibitors have an effect in terms of modulation of miRs and antitumour effects on different AML cell-lines harbouring mutations resembling those present in our patients.
A total of 23 patients aged 1 to 18 years, who received a diagnosis of AML harbouring FLT3-ITD or MLL rearrangement (Table SI). Although not mutually exclusive, these rearrangements are not frequently found together. In the present study, none of the patients had both the rearrangements. Bone marrow (BM) samples were collected from January 1st 2010 to December 31st 2016 at Bambino Gesù Children's Hospital in Rome and at Department of Paediatrics, University in Padua, at diagnosis and at disease recurrence from the 13 patients who underwent relapse (REL-D and REL-R groups, respectively) and at diagnosis from the 10 patients who did not display relapse (NREL group). A total of 8 frozen age-matched BM samples from healthy children (HD) (unused aliquots from healthy BM donors) were retrieved from the tissue bank at Bambino Gesù Children's Hospital as a control population. Informed consent was obtained from either parents or legal guardians according to the Declaration of Helsinki (2008). The present study was approved by the Institutional Review Boards of Bambino Gesù Children's Hospital (Rome, Italy).
Mononuclear cells were isolated by density gradient centrifugation at 400 g and 20°C for 30 min, diluted in 90% fetal bovine serum (FBS) plus 10% dimethyl sulfoxide (DMSO) and stored in liquid nitrogen. CD34+ cells from BM samples of three patients randomly selected in our cohort, were magnetically separated using MACS CD34+ microbead kit (Miltenyi Biotech GmbH). In particular, the molecular analysis of these patients revealed a FLT3-ITD with normal karyotype and two MLL rearrangements [t(9;11) and t(10;11)]. The identity of CD34 cells was validated by flow cytometry using FACSCantoII equipped with FACSDiva 6.1 CellQuest software (Becton, Dickinson and Company) using 20 µl of CD34 PerCP antibody (cat. no 340666; BD Biosciences) with an incubation of 30 min at 4°C.
Total RNA was extracted using TRIzol® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) and purified using RNA Cleanup and Concentration kit according to the manufacturer's protocol (Norgen Biotek Corp.). RNA quantification was performed using Nanodrop 2000 at 260 nm wavelength (Thermo Fisher Scientific, Inc.) and RNA integrity and purity was assessed with RNA Bioanalyzer kit according to the manufacturer's protocol (Agilent Technologies, Inc.). miR expression profile was performed using the nCounter Human v2 miRNA Expression Assay and nCounter Nanostring platform according to manufacturer's protocol (NanoString Technologies).
AML cell lines THP-1 (MLL-MLLT3; MLL-AF9), MOLM-13 (MLL-MLLT3; MLL-AF9; FLT3-ITD) and MV-4-11 (MLL-AFF1; MLL-AF4; FLT3-ITD) were obtained from DSMZ and cultured at 37°C using RPMI-1640 medium (Euroclone SpA) supplemented with 10% FBS (Thermo Fisher Scientific, Inc.) and 1% Penicillin-Streptomycin (Thermo Fisher Scientific, Inc.). Mycoplasma testing was performed for the cell lines used. Cell lines were treated with 250 nM JQ1 and 10 µM curcumin singularly alone or in association, for 48 h. JQ1 and Curcumin were obtained from Selleck Chemicals and resuspended in DMSO, following the manufacturer's protocol.
Whole-cell lysates were prepared with RIPA lysis buffer (Thermo Fisher Scientific, Inc.) supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific, Inc.). Cells were lysed by sonication, incubated for 30 min at 4°C and then obtained cells lysates were centrifuged at 13,000 × g for 20 min at 4°C. The protein concentration of the resulting supernatant was estimated by BCA assay. Then, 40 µg of sample was separated on Criterion TGX Precast Gels 4–20% (BioRad Laboratories, Inc.) and transferred to Hybond ECL nitrocellulose membranes (Amersham; Cytiva). Membranes were blocked at room temperature for 1 h in 5% non-fat milk in Tris buffered saline and 0,05% Tween-20 (TBS-T). Membranes were incubated at 4°C overnight with rabbit polyclonal anti-human CDKN1B (1:500; cat. no. sc-528; Santa Cruz Biotechnology, Inc.) and 1 h at room temperature with rabbit monoclonal anti-human GAPDH (1:1,000; cat. no. D16H11; Cell Signaling Technology, Inc.) primary antibodies. After incubation they were washed three times in TBS-T, then incubated with HRP-labelled goat anti-rabbit (1:5,000; cat. no. sc-2004) and goat anti-mouse (1:5,000; cat. no. sc-2005; both from Santa Cruz Biotechnology, Inc.) IgG secondary antibodies, respectively at room temperature for 1 h. Subsequently, they were washed an additional three times with TBS-T and then developed with ECL reagent (Western Lightning Plus; PerkinElmer, Inc.).
Expression levels of hsa-miR-221-5p, hsa-miR-222-5p and U6 were measured using TaqMan microRNA assays (cat. nos. 000524, 002276 and 001973; Thermo Fisher Scientific, Inc.). Reverse transcription (RT) primer, preformulated forward/reverse primer and MGB probes for each assay were provided by the manufacturer. The TaqMan MiR Reverse Transcription kit was used for cDNA synthesis from 10 ng total RNA template according to the manufacturer's protocol. QuantStudio 12K Flex Real Time PCR System (Thermo Fisher Scientific, Inc.) was used for qPCR reactions with the following conditions: Enzyme activation 95°C for 20 sec and 40 cycles of denaturation (95°C for 1 sec) and annealing/extension (60°C for 20 sec) steps. miRNA expression data were normalized to U6 using the 2−ΔΔCq (36) method by the Relative Quantification module of Thermo Fisher Cloud Data Analysis Apps. At least two independent amplifications were performed for each probe on triplicate samples.
Following treatment with 250 nM BRD4 and 10 µM curcumin, cells were washed twice with ice cold PBS and stained for 15 min at room temperature in calcium-binding buffer with PE-conjugated Annexin V (AnnV) and 7-Aminoactinomycin D (7-AAD) using the AnnV apoptosis detection kit (BD Pharmingen; BD Biosciences) according to the manufacturer's recommendations. Samples were analysed within 1 h by a fluorescence-activated cell sorting using a FACSCantoII equipped with FACSDiva 6.1 CellQuest software (Becton, Dickinson and Company).
MicroRNA profiling normalization was performed using the nSolver Analysis Software (NanoString Technologies) as recommended by NanoString. P-values were calculated using the LIMMA (v.3.46.0) package (37) from the Bioconductor R (v.4.0.5) project. The P-values were adjusted for multiple testing using the Benjamini and Hochberg method to control the False Discovery Rate. An independent normalization phase for each comparison was performed, considering only the samples present in such comparison (for example, NREL vs. HD). Then, the miRNA expression of the HD group was specifically normalized in the comparisons in which the HD group was taken into consideration. Validated targets of miRs were reported in Table I according to miRWalk 2.0 online software analysis (http://mirwalk.umm.uni-heidelberg.de/). One-way ANOVA and post hoc comparison using Tukey's HSD Post Hoc or Dunnett's test were performed using SPSS software v19 (IBM Corp.) and GraphPad Prism v6 (GraphPad Software, Inc.). The heatmap was generated by using GenePattern tool (38), with Euclidean and Spearman correlation distances in columns and rows, respectively. Venn diagrams were created using web tool (39).
CD34+ cells were cultured using MethoCult H4434 methylcellulose medium (Stem Cell Technologies) supplemented with 250 nM JQ1 and 10 µM curcumin.
A miR profiling analysis was first performed to verify whether the two distinct molecular subsets of the cohort of our patients (MLL rearranged and FLT3-ITD) showed different miR expression fingerprints. Comparing both MLL rearranged and FLT3-ITD sets with healthy donors (HDs), 4 and 16 significantly deregulated miRs were identified, respectively (Table I). miR-196b-5p and miR-34a-5p resulted upregulated in both AML sets. The comparison between the two molecular AML sets showed 3 differentially regulated miRs with miR-10a-5p and miR-99a-5p significantly higher in FLT3-ITD and miR-9a-5p with enhanced expression in MLL-rearranged sets (Table I). Other miRs such as miR-451a, miR-520d-5p/527/518a-5p, miR-574-5p and miR-192-5p were uniquely dysregulated in FLT3-ITD or MLL sets with respect to HDs (Table I). A miR expression profiling analysis was then performed based onto clinical outcomes of patients to identify those miRs associated with relapse. The hierarchical clustering results of miRNAs expression of REL-D vs. NREL groups is revealed in Fig. 1. A total of 18 miRs were broadly upregulated in the REL-D patients set with respect to the NREL set and 48 miRs displayed the opposite trend (Fig. 1A and Table II). To further identify and refine a signature that was associated with relapse, among these differentially expressed 66 miRs, only those shared in the REL-D vs. HD and NREL vs. HD comparison were subsequently considered (Fig. 1B and Table II). The resulted signature associated to relapse and not-relapse is listed in Table III. Validated targets of miRs were identified using miRWalk 2.0 online software (40) and are reported in Table III. CDKN1B, a key regulator of cell cycle which has been previously reported to be associated with prognosis in AML, resulted as primary target of miR-221-3p and miR-222-3p (41). To evaluate if the signature of miRs associated with relapse at diagnosis was maintained over time (if it is present also in the REL-R group), expression of miRs at diagnosis and at disease recurrence was compared between those patients who relapsed (REL-R vs. REL-D) and no significant differences were detected (Table II). Since our goal was to identify miRs associated to relapse with a significant prognostic value, the expression level of both miRs resulted overexpressed in REL-D vs. HD group and REL-D vs. NREL was first evaluated in the AML cells lines by qPCR (data not shown). High variability was obtained in the different cell lines. This result prompted the authors to focus only on hsa-miR-221-3p and hsa-miR-222-3p, presenting the same trend of expression in all the cell lines, as well as in vivo in the patients.
It was first analysed whether the combination of JQ1 and curcumin could modulate miR-221-3p and miR-222-3p expression in AML cell lines. RT-qPCR experiments (Fig. 2A), revealed a trend towards downregulation of the two miRs in all the cell lines, featuring an enhanced effect with the drug combination. The expression of miR-221-3p expression showed a clear trend towards downregulation, reaching statistical significance in MV-4-11 cells when treated with JQ1, curcumin or the combination of both (Fig. 2A, upper panel). miR-222-3p demonstrated a trend towards downregulation with coupled treatment in all cell-lines, but it did not reach statistical significance (Fig. 2A, lower panel). It was then analysed whether the combination of JQ1 and curcumin could modulate the CDKN1B protein level, that it is a miR-221-3p and miR-222-3p target, as before mentioned. As revealed in Fig. 2B, the treatment determined an increase in CDKN1B expression both in THP-1 and in MOLM-13 cells while in MV-411 cells the effect of drug combination on the upregulation of CDKN1B expression was comparable to that obtained by single treatment with JQ1.
It was assessed whether these treatments could drive the cell lines toward an apoptotic response. A significant increase in AnnV positive cell percentage was identified in all the cell lines while treated with the drug combination compared with DMSO (Figs. S1 and 3). At a deeper glance, THP-1 and MV-4-11 cells, showing a minor increase in CDKN1B expression, displayed a milder apoptotic response compared with MOLM-13, characterized instead by higher changes in term of expression of CDKN1B (Figs. 2 and 3).
miR-221-3p and miR-222-3p expression was analysed in CD34+ cells isolated from BM samples from three randomly selected patients from our cohort. The purity of CD34+ cells after isolation was 89% (Fig. S2). After CD34+ cells were cultured with MethoCult methylcellulose medium supplemented with JQ1 and curcumin, RNA was extracted and miR-221-3p, miR-222-3p expression was evaluated. In all the samples analysed miR-221-3p and miR-222-3p were modulated when treated with JQ1 and curcumin. In particular, the double treatment led to a significant downregulation of their expression (Fig. 4).
Despite improvements in the treatment of paediatric patients affected by AML, novel therapeutic strategies are needed, particularly for paediatric high-risk AML, characterized by high relapse incidence. miRs have been extensively studied as potential biomarkers in adult AML and, recently, expression signatures of miRs in paediatric samples have been proposed (14,19,42). Notably, few miRNA-based prognostic models have been proposed for paediatric cytogenetically normal AML (18,43), t(8;21) RUNX1-RUNX1T1 AML (44) and AML without considering the cytogenetics (18,45). However, conflicting results on the prognostic value of different miRs were reported, possibly due to the vast variability of miRs expression among different cytogenetic and molecular subtypes of AML (15). It was therefore decided, first, to characterize signature of miRs in patients harbouring FLT3-ITD and MLL rearrangement to verify whether distinct AML molecular subsets could affect different miR expression fingerprints. Moreover, to further narrow our miRs signature (identifying only AML associated miRs), HDs were used as a control group. Data on expression of miRs among different AML molecular subtypes are available (16). In accordance with literature data (46), miR-9-5p and miR-10a-5p were upregulated in MLL-rearranged and FLT3-ITD sets, respectively, even compared with HDs. In addition, miR-99a-5p was upregulated in FLT3-ITD patients as compared with MLL patients, and was already described in literature as a potential oncomiR in paediatric AML (42). miR-196b-5p and miR-34a-5p were both upregulated in MLL and FLT3-ITD with respect to HDs and they are possibly involved in the pathogenesis of both AML variants. miR-34a is a widely studied miR as a promising target and it has been considered as a preclinical and clinical model for the treatment of solid tumours, myeloma and B-cell lymphoma (47,48). Despite the limited number of patients, to the best of our knowledge, this is the first study investigating the role of miRs as predictive biomarker of relapse in paediatric AML patients with FLT3-ITD or MLL rearrangement. A total of 28 miRs differentially expressed were identified between patients who relapsed and those who did not, being possibly associated with prognosis. Among them, the vast majority of miRs have already been reported as related to cancer. Comparing the signature of our miRs to those previously reported, a common dysregulation of mir-155 was identified, but displaying an opposite behaviour with a protective effect in the present study as opposed to a relapse association in others (19,49). miR-155 has led to conflicting results in previous studies, with certain groups reporting an anti-leukemic role (50) and others (13,51–53) reporting a role as oncomiR in AML. This is possibly explained by a different level of expression of miR-155, acting as a tumour suppressor when highly expressed and as an oncomiR when overexpressed to an intermediate level (54). miR-34a-5p expression was positively correlated with relapse. It was already described that patients with low miR-34a expression showed shorter overall and recurrence-free survival (55). Numerous of the identified miRs, have also been independently considered as promising tool to develop novel therapies; these include miR-200c (47,56), which was revealed to be associated to NREL patients. miR-221-3p and miR-222-3p, both associated to relapse, gained the attention of the authors as they have been broadly reported in literature as oncomiRs both in hematologic malignancies such as chronic lymphocytic leukemia (57), myelodisplastic syndrome (58), acute lymphoblastic leukemia (59) and AML (60,61) as well as in solid tumours. These miRs have as their primary target CDKN1B, a master regulator of the cell cycle. CDKN1B is a well-known cyclin-dependent kinase inhibitor which regulates cell cycle progression at G1 stage, preventing the activation of cyclin E-CDK2 or cyclin D-CDK4 complexes, resulting in a blockade of cell division cycle (62). BET and HAT inhibitors have already been associated with modulation of miRs in hematological malignances (63,64). The present findings revealed, for the first time, that JQ1 determines a clear trend towards downregulation of miR-221-3p expression in both MOLM-13 and MV-4-11AML cell lines with a synergic effect when associated with curcumin; reaching statistical significance in the second one. This is interesting also considering that for FLT3-ITD rearrangement MOLM-13 expresses both mutated and wild-type allele, while MV-4-11 expresses mutated allele only (65). It could be hypothesized that FLT3-ITD in both mutated alleles renders MV-4-11 cells more sensitive to the effect of drug on miR-221-3p modulation. Moreover, following the combined treatment, an increase was demonstrated in CDKN1B expression and in apoptotic response in our AML cell lines. The present results were confirmed in cultures with primary leukemic cells, showing a marked reduction of miR-221-3p and miR-222-3p expression. These results supported the idea that BET inhibitors, along with curcumin, could regulate not only coding RNA transcription, but also non-coding RNA such as miRs. Although the combination of JQ1 and curcumin synergistically reduced miR-221 and miR-222 expression and increased apoptosis in AML cells, a limitation to the present study was represented by insufficient patient samples. Direct regulation of CDKN1B by miR-221-3p and miR-222-3p should be confirmed by further experiments including western blot analysis to verify expression levels of CDKN1B in samples of patients and HDs and the use of inhibitors of miR-221 or miR-222 in a bigger cohort of patients. In conclusion, the present study identified fingerprints of miRs related to relapse and non-relapse in paediatric FLT3-ITD- or MLL-rearranged AML. Numerous of these miRs are known to be involved in pathogenetic mechanisms of several haematological malignancies as well as solid tumours and represent both good candidates for targeted treatments and therapeutic tools in different neoplasms. The use of the well-known BRD4 inhibitor JQ1, as well as novel BRD inhibitors in the care of leukaemia could be potentiated by epigenetic drugs such as HATs inhibitors, antagomiR or miR mimic and could expand the therapeutic arsenal in HR-AMLs, particularly for paediatric patients. |
PMC9647507 | Christine E. Robbins,Bhumil Patel,Danielle L. Sawyer,Barrie Wilkinson,Brian K. Kennedy,Mark A. McCormick | Cytosolic and mitochondrial tRNA synthetase inhibitors increase lifespan in a GCN4/atf-4-dependent manner | 21-10-2022 | Biological sciences,Biochemistry,Molecular biology | Summary Deletion of genes encoding ribosomal proteins extends lifespan in yeast. This increases translation of the functionally conserved transcription factor Gcn4, and lifespan extension in these mutants is GCN4-dependent. Gcn4 is also translationally upregulated by uncharged tRNAs, as are its Caenorhabditiselegans and mammalian functional orthologs. Here, we show that cytosolic tRNA synthetase inhibitors upregulate Gcn4 translation and extend yeast lifespan in a Gcn4-dependent manner. This cytosolic tRNA synthetase inhibitor is also able to extend the lifespan of C. elegans in an atf-4-dependent manner. We show that mitochondrial tRNA synthetase inhibitors greatly extend the lifespan of C. elegans, and this depends on atf-4. This suggests that perturbations of both cytosolic and mitochondrial translation may act in part via the same downstream pathway. These findings establish GCN4 orthologs as conserved longevity factors and, as long-lived mice exhibit elevated ATF4, leave open the possibility that tRNA synthetase inhibitors could also extend lifespan in mammals. | Cytosolic and mitochondrial tRNA synthetase inhibitors increase lifespan in a GCN4/atf-4-dependent manner
Deletion of genes encoding ribosomal proteins extends lifespan in yeast. This increases translation of the functionally conserved transcription factor Gcn4, and lifespan extension in these mutants is GCN4-dependent. Gcn4 is also translationally upregulated by uncharged tRNAs, as are its Caenorhabditiselegans and mammalian functional orthologs. Here, we show that cytosolic tRNA synthetase inhibitors upregulate Gcn4 translation and extend yeast lifespan in a Gcn4-dependent manner. This cytosolic tRNA synthetase inhibitor is also able to extend the lifespan of C. elegans in an atf-4-dependent manner. We show that mitochondrial tRNA synthetase inhibitors greatly extend the lifespan of C. elegans, and this depends on atf-4. This suggests that perturbations of both cytosolic and mitochondrial translation may act in part via the same downstream pathway. These findings establish GCN4 orthologs as conserved longevity factors and, as long-lived mice exhibit elevated ATF4, leave open the possibility that tRNA synthetase inhibitors could also extend lifespan in mammals.
Aging is known to be a modifiable phenotype. Multiple pathways have been identified that are conserved from nonvertebrate models through mice (Blüher et al., 2003; Holzenberger et al., 2003; Kennedy et al., 1995; Kenyon et al., 1993; Klass, 1983; McCormick et al., 2015), and these have now led to drug trials in dogs (Urfer et al., 2017) and humans (Mannick et al., 2014). Furthermore, many model organisms exhibiting increased lifespan also show delayed onset of many phenotypes and diseases of aging (Garigan et al., 2002; Martin-Montalvo et al., 2013; Vora et al., 2013), suggesting that these conserved interventions extend the healthy disease-free period of life, or healthspan, in humans. In our previously completed genome-wide screen for increased replicative lifespan (RLS) in the budding yeast Saccharomyces cerevisiae, we identified many deletions of ribosomal protein-encoding genes and found that this category of genes is statistically overrepresented among deletions that increase the RLS of yeast (McCormick et al., 2015; Steffen et al., 2008). These ribosomal protein gene deletions lead to increased translation of the nutrient-responsive yeast transcription factor Gcn4, and the increased RLS of these strains is mostly suppressed by the deletion of GCN4 (Steffen et al., 2008, 2012). The Gcn4 pathway is functionally conserved from yeast through humans. In Caenorhabditis elegans, a very successful model organism for the study of aging, the functional ortholog of yeast GCN4 is atf-4, which until recently was referred to as atf-5, and in mammals, it is ATF4. In addition to being upregulated by the deletion of yeast ribosomal proteins, Gcn4 translation can also be upregulated by the accumulation of uncharged tRNAs (Krupitza and Thireos, 1990; Wek et al., 1989). tRNA Synthetases, which promote the association of an amino acid to the appropriate tRNA, are highly conserved essential proteins, and as such, several species have evolved to form highly specific inhibitors of tRNA synthetase activity (O’Donoghue and Luthey-Schulten, 2003). We reasoned that it might be possible to identify a dose range for these compounds that did not appreciably inhibit growth or fitness by completely blocking translation but still led to an accumulation of uncharged tRNAs. This could potentially increase Gcn4 translation and recapitulate the increased lifespan we saw in ribosomal protein-deletion mutants with increased Gcn4. If this were the case, we would expect any increase in lifespan to be GCN4/atf-4/ATF4 dependent. Interestingly, work in mice has now shown that multiple types of long-lived mice show increased ATF4 levels (Li and Miller, 2015; Li et al., 2014). Thus, compounds that increased the lifespan in invertebrates by upregulating the Gcn4 pathway might lead to drug treatments that could extend the lifespan in mammals. We identified a biologically derived tRNA synthetase inhibitor, borrelidin, that leads to large increases in Gcn4 translation that are completely dependent on the uncharged tRNA sensor Gcn2. Borrelidin extends yeast RLS specifically at highly Gcn4-inducing doses, and this increase in lifespan depends completely on the presence of GCN4. We then asked whether this phenotype might be conserved and whether uncharged tRNAs generated by specific disruption of mitochondrial translation might trigger this response. Along with borrelidin, we identified another mitochondrial tRNA synthetase inhibitor, mupirocin, that greatly extends the lifespan of C. elegans in a dose-dependent manner. We found that this increased lifespan due to both these inhibitors is completely dependent on the presence of atf-4, the C. elegans functional ortholog of GCN4.
Borrelidin is a biologically synthesized polyketide that inhibits threonyl tRNA synthetase. It is produced by several Streptomyces species and was first isolated from Streptomyces rochei (Berger et al., 1949). We reasoned that at a low-enough dose to prevent lethality, an appropriate tRNA synthetase inhibitor might still lead to accumulation of uncharged tRNAs, thus activating Gcn2, leading to phosphorylation of Sui2 (yeast eIF2α), and thus to increased translation of Gcn4 (Hinnebusch, 2005; Krupitza and Thireos, 1990), as outlined in Figure 1A. In yeast, doubling time, or the mean time for a population to double in number, is a direct and commonly used measure of growth rate. Specifically, during the exponential (“log”) growth phase in yeast, where N0 = Nt ∗e(gr∗t) t, (N0, number of yeast at time zero, Nt, number of yeast at time t), the doubling time Td and the growth rate gr are related by the formula Td = ln(2)/gr. We first measured the effects of borrelidin on the yeast growth rate, quantified in terms of doubling time, for yeast grown at a wide range of borrelidin concentrations in a standard liquid media (Yeast Extract Peptone +2% glucose at 30°C). The results, shown in Figure 1B, indicate that a borrelidin concentration of 80 μM is able to greatly increase the yeast doubling time without completely stopping growth, while concentrations below 20 μM have no significant effect on the yeast doubling time. This suggests that concentrations above 20 μM should be biologically active for yeast grown in liquid media. Using these data as a starting estimate of the biologically active range of borrelidin concentrations in yeast liquid media, we then measured yeast growth on solid media (YEP agar +2% glucose at 30°C), at a range of borrelidin concentrations, finding that concentrations from 40 to 320 μM were sufficient to greatly inhibit growth, while concentrations below 20 μM did not appreciably slow growth. We repeated these experiments on solid media kept at both 4°C and 30°C for 15 days in order to confirm that the biological activity of borrelidin was not lost under conditions of yeast RLS experiments. Growth on solid media in all cases tested is shown in Figure 1C. The increased amounts of borrelidin needed to produce the same growth inhibition on solid media presumably reflect a less-efficient uptake of the drug by yeast under these conditions. Additionally, our results agree with a large-scale review of drug screening in S. cerevisiae with diverse compounds, in both solid and liquid media, in which the authors noted that solid media-based assays typically require 1–2 orders of magnitude higher drug concentrations than liquid (Smith et al., 2010).
Once we had identified biologically active doses of borrelidin in both solid and liquid media, we asked whether these doses could induce translation of Gcn4. Gcn4 is regulated at the level of translation by Gcn2/eIF2 (and thus by uncharged tRNA accumulation), through the activity of upstream open reading frames (uORFs) in the 5′ untranslated region of the GCN4 mRNA (Mueller and Hinnebusch, 1986), and this mechanism of translational regulation is conserved through mammals (Vattem and Wek, 2004). We measured Gcn4 translation using a dual-luciferase reporter (Steffen et al., 2008) with the GCN4 promoter, GCN4 5′ untranslated region, and GCN4 ORF fused in-frame to firefly luciferase, with the PGK1 (yeast 3-phosphoglycerate kinase) promoter driving Renilla luciferase as an on-plasmid control, as illustrated in Figure 2A. In liquid media, concentrations of borrelidin from 0.16 to 5 μM showed increasing induction of Gcn4 up to 5-fold over control and then decreased as the concentrations reached those that inhibit yeast growth in liquid media, from 10 to 80 μM, as illustrated in Figure 2B showing fold induction relative to DMSO-only control. In parallel, we measured the effect of borrelidin on gcn2Δ yeast, finding this strain refractory to induction of Gcn4 (Figure 2C), as predicted by the pathway outlined in Figure 1A. This pathway also suggests that increased Gcn4 translation should depend on SUI2 (yeast eIF2α), but this could not be tested directly as SUI2 is an essential gene in yeast (Cigan et al., 1989). We next asked which doses of borrelidin could induce increased Gcn4 translation of yeast grown on solid media, and we found that roughly 8-fold higher doses were required to achieve the same increased translation of Gcn4 relative to liquid media, as observed in the case of growth inhibition (Figure 2D). We determined whether the Gcn4-inducing activity of borrelidin in solid media was decreased after 15 days at 30°C, in order to estimate whether this activity would be likely to be maintained over the course of a yeast RLS experiment. We found that there was no loss in Gcn4 induction by borrelidin in solid media after 2 weeks at 30°C (Figure 2E), again shown as fold induction relative to DMSO-only control. While our dual-luciferase reporter will reflect any changes in both transcription and translation of GCN4, GCN2 is only known to act in the regulation of the translation of GCN4, not transcription. Finally, we measured the overall effects of borrelidin on protein synthesis in both wild-type and gcn4Δ yeast (Figures 2F and 2G). While there is a predictable dose-dependent decrease in protein synthesis in wild-type yeast treated with the tRNA synthetase inhibitor, this effect is not dependent on GCN4. Taken together, these data further suggest that the observed changes in GCN4 are due to changes specifically in translation, which are induced at certain biologically active concentrations of the tRNA synthetase inhibitor, borrelidin via Gcn2.)
After identifying borrelidin concentrations that increased Gcn4 translation in yeast grown on solid media and that maintained this activity after 2 weeks at 30°C, we asked whether these concentrations could increase yeast RLS, as was seen previously in ribosomal protein deletion strains that induce Gcn4 translation (Steffen et al., 2008, 2012). As shown in Figure 3A, yeast RLS was increased by borrelidin, at a highly Gcn4-inducing dose as shown in Figure 2D. Perhaps unsurprisingly, very high doses of borrelidin that greatly inhibit yeast growth showed decreased RLS (Figure S1A, related to Figure 3). We also asked whether the increased RLS seen in borrelidin-treated yeast required GCN4, as predicted by the pathway outlined in Figure 1A. Figure 3B shows that in gcn4Δ yeast, borrelidin treatment does not increase yeast RLS at all, and in fact, it slightly shortens it. At very high doses of borrelidin that greatly inhibit yeast growth, RLS is shortened in gcn4Δ yeast exactly as it is in wild-type (Figure S1B, related to Figure 3). This suggests that while the extended lifespan at lower doses of borrelidin due to the accumulation of uncharged tRNAs is dependent on GCN4, the shortened lifespan at much higher growth-inhibiting concentrations is a consequence of blocked overall translation and does not act through Gcn4. These data fit a model where tRNA synthetase inhibitor-induced lifespan extension is the result of uncharged tRNAs activating GCN2-eIF2alpha-GCN4 signaling in a GCN4-dependent manner. However, an excessive accumulation of uncharged tRNA can interfere with overall translation in a GCN4-independent manner that does not act through GCN2 or eIF2alpha. Furthermore, these data suggest that any lifespan effects of tRNA synthetase inhibition are likely to be extremely dose dependent. We wanted to determine if this lifespan extension by borrelidin could be genetically modeled. As deletion of THS1 is lethal, we utilized the Tet-Off system that offers a tunable reduction of gene expression by the addition of varying concentrations of doxycycline (Mnaimneh et al., 2004). The yeast Tet-promoters collection offers two different THS1 strains. Both strains have significantly increased lifespan when treated with doxycycline while the control R1158 strain does not (Figure S2, related to Figure 3). While borrelidin is capable of having other biological effects such as roles in the unfolded protein response, synthesis of nitrile, and angiogenesis in mammals (Wakabayashi et al., 1997; Wang et al., 2015), the GCN2 dependence of the Gcn4 upregulation, and GCN4 dependence of the lifespan extension, leads us to favor a model whereby the tRNA synthetase inhibitor activity of borrelidin is key to the increased lifespan we have observed.
Previous work has shown that depletion of C. elegans mitochondrial ribosomal subunits by RNAi knockdown increases lifespan (Hansen et al., 2007; Houtkooper et al., 2013), and very recently, lifespan extension by knockdown of one of these, mrps-5, has been shown to partially depend on atf-4 (Molenaars et al., 2020), providing further evidence of a conserved mechanism. Based on this previous work, we wanted to ask whether the phenotypes we had observed could be recapitulated by the inhibition of mitochondria-specific tRNA synthetases. Mitochondrial translation and mitochondrial tRNA synthetase activity have been recently thoroughly reviewed (D’Souza and Minczuk, 2018), but in brief, the mitochondria is responsible for only translating a small number of mitochondrial proteins including mRNAs encoding proteins required for oxidative phosphorylation. While the mitochondrial genome encodes the mitochondrial tRNAs, the mitochondrial tRNA synthetases responsible for charging tRNAs with their cognate amino acid are coded in the nucleus and transported to the mitochondria. Other proteins, such as initiation factors, required for mitochondrial translation are also encoded in the nucleus and transported to the mitochondria. While mitochondrial and cytosolic tRNA perform the same job in different cellular regions, the mitochondrial amino acid code that the charged tRNA recognizes differs from the universal cytosolic code. Mupirocin is a biologically synthesized inhibitor of isoleucyl tRNA synthetase initially isolated from Pseudomonas fluorescens that shows an 8000-fold greater affinity for prokaryotic Ile-tRNA synthetase over the eukaryotic one (Fuller et al., 1971; Hughes and Mellows, 1980). As a result, the effects in C. elegans should be mediated by inhibition of the C. elegans mitochondrial isoleucyl tRNA synthetase, iars-2, rather than the cytoplasmic isoleucyl tRNA synthetase, iars-1. We first assayed the biological activity of mupirocin on C. elegans by exposing multiple C. elegans eggs in liquid culture to varying concentrations of mupirocin. At all concentrations, eggs developed normally into healthy adults, but they did so at varying rates. Doses below 0.04 mM showed no discernible effect, while doses between 0.1 mM and 5 mM showed increasingly slower development through all larval stages, without any noticeable growth arrest (Figure 4A). Development was measured every 24 h for 10 worms per well per day until adulthood was reached. There was no discernible variation in development within wells. Interestingly, these developmental delays are completely independent of the presence of worm atf-4 (Figure 4B), suggesting that they are a consequence of lowered overall translation during the rapid growth phase of C. elegans development due to tRNA synthetase inhibition and not downstream of the Gcn4/ATF-4/ATF4 pathway signaling. Having confirmed that there was a range of doses that were biologically active on C. elegans in liquid culture, we turned next to solid culture, as this is the media we use in lifespan assays. We found that, again, mupirocin treatment led to delays in development without arrest consistent with liquid developmental delay data, with no discernible delay for doses up to 0.04 mM, and increasingly slowed development for doses from 0.15 to 3.08 mM. This was measured for an average of 34 worms per strain and mupirocin concentration, spread onto four plates, with almost zero variability observed both within and between plates for all worms at a given strain and concentration. Quantification of times to reach adulthood from egg on solid media for wild-type worms at various concentrations of mupirocin is shown in Figure 4C. As with liquid culture, all developmental delays were completely unchanged in worms deleted for atf-4, specifically atf-4(ok576), as shown in Figure 4D. Representative growth images for wild-type N2 worms on solid media after 72 h at 20°C are shown in Figure 4E, and representative growth images for atf-4(ok576) worms on solid media after 72 h at 20°C are shown in Figure 4F. As mupirocin is thought to act on the mitochondrial Ile-tRNA synthetase, we investigated the effects it has on the development of the already developmentally delayed mitochondrial electron transport chain mutant, isp-1(qm150) (Feng et al., 2001). As seen in Figure S3, related to Figure 4, the developmental delay of mupirocin appears to be additive to the mitochondrial mutant’s developmental delay. At the highest concentrations of mupirocin, not only was this additional developmental delay less consistently dose dependent but many animals failed to develop into adults at all. Taken together, these data show that there is a consistent delay in development in worms treated with tRNA synthetase inhibitors that are independent of atf-4 activity.
Having confirmed that there was a range of doses where mupirocin was biologically active in worms in both liquid and solid media, we then turned to lifespan. All worm lifespans were scored as survival beginning from day 1 of adulthood for that specific cohort of worms, so that any developmental delays did not influence survival times. We observed that mupirocin treatment led to a dramatic increase in mean and maximum lifespan, with increasing doses of mupirocin leading to increasing lifespan. At no concentrations did we observe any ill effects. This suggests that it is possible that a higher effective dose of tRNA synthetase inhibitor, perhaps through a different route of administration, could possibly increase lifespan even further. Survival curves for worms at increasing concentrations of mupirocin are shown in Figure 5A, and the relationship between dose and mean lifespan is shown in Figure 5B. Significant increases in lifespan in Figure 5A are interpreted relative to the 0-mM control that contains only a DMSO vehicle. This is because DMSO itself (the solvent for mupirocin in these experiments) has previously been shown to more modestly, but significantly, extend lifespan in C. elegans (Frankowski et al., 2013). Survival curves and mean lifespan vs. dose plots also containing the no DMSO and no drug control are shown in Figure S4, related to Figure 5. Next, we asked whether the increased lifespan seen upon administration of tRNA synthetase inhibitors depended on the presence of the C. elegans functional ortholog of GCN4, atf-4. We tested the effects of mupirocin on lifespan in atf-4(ok576), an atf-4 deletion mutant generated by the C. elegans Deletion Mutant Consortium (C. elegans Deletion Mutant Consortium, 2012). No increase in lifespan was seen at any high concentration of mupirocin in atf-4 deletion mutants, as shown in survival curves in Figure 5C and the dose vs. mean lifespan plot in Figure 5D. Additionally, we measured the lifespan of the mitochondrial electron transport chain mutant, isp-1(qm150), treated with mupirocin. As seen in Figure S5A, related to Figure 5, the lifespan extension of isp-1 mutants is not increased when treated with mupirocin and was lethal at higher concentrations This is unlike the additive effects seen for developmental delay. We also examined the lifespan effects of mupirocin on a gcn-2 deletion strain, gcn-2(ok886). As in yeast, this gene functions as one of the upstream regulators of ATF-4/Gcn4 activity. As seen in Figure S5B, related to Figure 5, at all doses of mupirocin, there is a reduced lifespan extension in gcn-2(ok886) worms compared to wild type, and any extension seen is only significant at the highest concentrations. This slight but significant difference is not seen in atf-4(ok576) worms treated with mupirocin. One potential explanation for this increase is that much of the lifespan extension occurs in a gcn-2-mediated manner, but there may be an additional mechanism involved that acts at higher concentrations in an atf-4-dependent but gcn-2-independent manner. As the C. elegans mitochondrial isoleucyl tRNA synthetase iars-2 is the predicted target of mupirocin, we asked whether RNAi knockdown of iars-2 had any effect on lifespan in wild-type or atf-4(ok576) worms and found that it did not (Figure S6, related to Figure 5). We hypothesize that there is a narrow beneficial range of tRNA synthetase activity reduction, where atf-4-dependent mechanisms are activated yet not offset by deleterious effects of overall lowered translation. It is possible that the reduction of iars-2 activity by RNAi using 1-mM isopropyl β-D-1-thiogalactopyranoside for induction did not induce a level of reduction in this beneficial range. Another possible explanation is that another gene such as iars-1 is able to compensate for the loss of iars-2 activity. In addition to mupirocin, we found that borrelidin is able to significantly extend the lifespan of worms at lower concentrations in a dose-dependent manner (Figure 5E). This lifespan extension is also dependent on atf-4 (Figure 5F). Again, RNAi knockdown of tars-1, the target of borrelidin, is unable to mimic this effect (Figure S6, related to Figure 5). This shows that lifespan extension through drug-induced uncharged cytosolic and mitochondrial tRNA is dependent on atf-4. We have found that mupirocin is unable to increase lifespan in yeast in glucose- or glycerol-containing media (Figure S7, related to Figure 5). We propose that this could be due to the mitochondrial-specific nature of the drug. As yeast do not require the use of their mitochondria for fermentation, it is possible that a drug that primarily effects a mitochondrial tRNA synthetase might be less likely to lead to a phenotype in yeast.
As mupirocin has the most significant lifespan extension, we determined the life stages during which mupirocin predominantly exerts its lifespan-extending effects in worms. This was in part due to the effects of mupirocin on developmental timing although these were clearly shown to be completely independent of atf-4. While the lifespan shown in Figures 5A and 5C depicts treated worms from hatching until death, we next compared treatments at different times. One condition, adult-only, involved eggs hatched and grown to adulthood on solid media with no drug and then transferred at adulthood to plates containing varying concentrations of mupirocin. The other, development-only, involved eggs grown on varying concentrations of mupirocin from hatching to adulthood and then transferred to plates containing no drug from then until death. The survival curves for the adult-only mupirocin treatment are shown in Figure 6A, and those for development-only mupirocin in Figure 6B. As before, lifespan differences were compared to the 0-mM DMSO-only control, but complete data with no drug and no DMSO control can also be found in Figure S8, related to Figure 6. These results suggest that the lifespan-extending effects of mupirocin treatment are largely separable from the developmental delays, as adult-only treatment extends lifespan less than whole-life treatment, and development-only treatment extends lifespan still less than adult-only treatment. The fact that adult-only treatment can clearly still extend lifespan again supports a model in which the developmental delay effect of mupirocin is separable from its effect on lifespan. At the same time, we propose that the decreasing effect on lifespan seen by whole-life treatment > adult-only treatment > development-only treatment can be explained by the relative times that worms are exposed to mupirocin in these three treatments: longest for whole-life, then for adult-only, and shortest for development-only.
Aging is a common underlying risk factor for many of the most significant causes of mortality and morbidity in most countries, such as Alzheimer’s disease, cancer, and heart disease. By studying pathways that can delay aging, we may be able to eventually delay the onset of these and other diseases and perhaps uncover the underlying causes of aging itself. Of the pathways now known to greatly affect aging, several, such as TOR signaling (Jia et al., 2004; Kaeberlein et al., 2005; Kapahi et al., 2004; Powers et al., 2006), insulin/IGF-1 signaling (Blüher et al., 2003; Clancy et al., 2001; Holzenberger et al., 2003; Kenyon et al., 1993), and chemosensation of food (Alcedo and Kenyon, 2004; Libert et al., 2007) involve nutrient sensing. Gcn4 is a nutrient-sensing protein that responds in part to nitrogen starvation, which can be upregulated by accumulation of uncharged tRNAs (Krupitza and Thireos, 1990; Wek et al., 1989) or by the deletion of cytosolic ribosomal proteins in yeast (Steffen et al., 2008, 2012). Mutations in genes encoding proteins in the cytosolic ribosome have also been shown to extend lifespan in C. elegans (Curran and Ruvkun, 2007; Hansen et al., 2007; Reis-Rodrigues et al., 2012). The data presented here suggest a model (Figure 7) in which accumulation of uncharged tRNAs, whether due to inhibition of mitochondria-specific tRNA synthetases or cytosolic ribosome tRNA synthetases, can lead to activation of the Gcn4/ATF-4/ATF4 nutrient-sensing pathway, and from there to increased lifespan in multiple organisms. Another group has recently shown that inhibition of the integrated stress response extends lifespan in C. elegans (Derisbourg et al., 2021); however, we have clearly shown that the lifespan extension caused by tRNA synthetase inhibitors is dependent on the activity of this very pathway. While these compounds can clearly have an effect on other proteins, we have shown a clear Gcn4 dependence on lifespan extension. It is likely that the inhibition of tRNA synthetases will affect the translation of other proteins, and a search for epistatic modifiers could identify other translational changes that are important to the observed increased lifespan. It makes sense that cells might interpret uncharged tRNAs as a sign of amino acid starvation and translational stress, whether these arise from problems with cytosolic translation or with mitochondrial translation. Indeed, given the mechanism of uncharged tRNA sensing by Gcn2 and its orthologs (Krupitza and Thireos, 1990; Wek et al., 1989), it is hard to imagine how this sensor protein can distinguish tRNAs from these two different origins. This suggests more broadly that uncharged tRNAs may signal mitochondrial translation defects and that some types of translation stress can lead to the same transcriptional and cellular responses, whether they are cytosolic or mitochondrial in origin. Other work has also indicated methods of tRNA transport including export from the mitochondria to the cytosol where Gcn2 and its orthologs may be able to encounter uncharged mitochondrial tRNA (Maniataki and Mourelatos, 2005). These transport mechanisms may be responsible for the activation of Gcn2 and upregulation of Gcn4 from uncharged mitochondrial tRNA, but this will need further investigation. While our current model is that mupirocin only acts as an inhibitor of mitochondrial Ile-tRNA synthetase as it shows a strong preference for prokaryotic tRNA synthetase, we have not ruled out the possibility that mupirocin could also act on the cytosolic Ile-tRNA synthetase, or even on other previously undiscovered targets. Previous work supports our model as mupirocin-resistant parasites developed mutations in their organelle-encoded tRNA synthetases rather than their cytosolic forms (Istvan et al., 2011). Further studies will be needed to discover any additional activities of this drug. We have shown by our use of pharmacological agents that there is a fine-tuned concentration in order to achieve the optimal effect and observed changes in lifespan without halting translation altogether or to a deleterious extent. Deletion of many tRNA synthetases in yeast and worms is lethal, and we have found that the use of the Tet-off system in yeast or RNAi in worms is not capable of eliciting the same effect in our hands and is perhaps not sensitive enough to mimic the precise effect. Other work has also shown that tRNA synthetase RNAi knockdown is able to extend lifespan in C. elegans, particularly under starvation conditions (Webster et al., 2017), or in the case of mitochondrial tRNA synthetase RNAi, it is able to impair mitochondrial function and extend lifespan (Lee et al., 2003). Others have shown RNAi knockdown of methionyl-tRNA synthetase can increase lifespan in Drosophila (Suh et al., 2020), further suggesting a conserved mechanism of lifespan extension. While inhibition of both cytosolic and mitochondrial tRNA synthetases using mupirocin and borrelidin is able to significantly extend lifespan in C. elegans, only borrelidin has been able to increase yeast-RLS under conditions we have tested. We show that both compounds extend lifespan through Gcn4/ATF-4, but the differences in drug activity will need to be further investigated. The differences between single and multicellular organisms are an obvious potential cause of these observed reactions. Yeast and worms also differ in their need for mitochondrial respiratory function. Yeast is capable of surviving without the respiratory function of their mitochondria, and this biological quirk could also be responsible for the lack of mupirocin effects seen in yeast. These findings have broader impacts, as several longevity-inducing interventions in mammals are associated with increased levels of the mammalian Gcn4/ATF-4 ortholog ATF4 (Li and Miller, 2015; Li et al., 2014). Given that many antibiotics target mitochondrial translation and are well-tolerated, these findings suggest new strategies for interventions to impact lifespan and healthspan. Moreover, they call for a more complete understanding of the role of ATF4 in mammalian aging.
Our interpretation of the results reported here makes the assumption that for both borrelidin and mupirocin, the phenotypes we have observed are primarily due to their activities as inhibitors of tRNA synthetase activity. While some of the genetic experiments, such as dependence specifically on the uncharged tRNA-sensing kinase GCN2/gcn-2, are supportive of this interpretation, we have not directly demonstrated any change in levels of charged and uncharged tRNA upon drug treatment. This leaves open the possibility that rather than tRNA synthetase inhibition, other previously reported targets or activities of these two drugs, which we have drawn attention in the main text, or even other previously unknown targets or activities could be responsible for the phenotypes we have observed.
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Mark McCormick ([email protected]).
No new reagents were generated in this study. All worm strains used are available from the Caenorhabditis Genetics Center https://cgc.umn.edu/. E. coli OP-50 is available from the Caenorhabditis genetics center https://cgc.umn.edu/. HT115, iars-2 RNA1, and tars-1 RNAi E. coli are available in the Ahringer RNAi library from Horizon Discovery, Boulder, CO, USA. All yeast strains are included in either the yeast deletion collection available from Horizon Discovery, Boulder, CO, USA, or the yeast tet-off promoter collection available from Horizon Discovery, Boulder, CO, USA.
All yeast strains were derived from the parent strains of the haploid yeast ORF deletion collections BY4742 (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) and BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0),with further deletions (e.g. gcn4Δ or gcn2Δ) marked with the KanMX cassette (Winzeler et al., 1999). The entire collection, as well as individual wild type, gcn4Δ, and gcn2Δ strains in both mating types, are available directly from Horizon Discovery, Boulder, CO, USA. Cells were grown on standard YPD containing 1% yeast extract, 2% peptone and 2% glucose, at 30°C, unless otherwise stated. All worm strains were maintained at 20°C on NGM plates containing E. coli strain OP-50 as a food source (Brenner, 1974). Wild-type N2 worms, atf-4 deletion worms RB790 atf-4(ok576), isp-1 mutant worms MQ887 isp-1(qm150), and gcn-2 deletion worms RB980 gcn-2(ok886) as well as E. coli strain OP-50 are all available directly from the Caenorhabditis Genetics Center https://cgc.umn.edu/.
Yeast RLS assays were performed as described in (Kaeberlein et al., 2004; Steffen et al., 2009). Specifically, virgin daughter cells were isolated from each strain and then allowed to grow into mother cells on YPD (yeast extract peptone dextrose) or other media as specified. These cells were microdissected each day in round intervals corresponding to the cell division time at that stage of the RLS (yeast divide more slowly as they age), while grown at 30°C in a Thermo/Thelco 6DG laboratory incubator. These yeast were transferred to 4°C overnight each day. This was done 7 days a week, continuously for the entire RLS, using a 40–1600X Trinocular Infinity-corrected Microscope with LED Koehler Illumination (T620A, AmScope, Irvine, CA) and an attached microdissection apparatus (Tetrad Manipulator AxioScope A1,w/Stage, Carl Zeiss Microscopy, Thornwood, NY), with attached 50 micron optical fiber dissection needles (1050, Styles Lab Supplies, Talent, OR). For each mother cell the corresponding daughters were microdissected and counted, until the mother cell could no longer divide.
All yeast growth rates in liquid media were analyzed in the OD420-580 range in the Bioscreen C automated microbiology growth curve analysis system (Growth Curves USA, Piscataway, NJ, USA) using the Yeast Outgrowth Data Analyzer (YODA) software (Olsen et al., 2010). For growth on solid media, 50μL of yeast at an OD600 of 1 were plated onto 1 mL of YEP agar +2% glucose per well in 24-well plates (CLS3738, Corning, Tewksbury, MA, USA) and growth was photographed every 24h for 120h. Images were quantified using FIJI/ImageJ (Schindelin et al., 2012).
Gcn4 translation was measured using a dual-luciferase reporter plasmid pVW31 (Steffen et al., 2008). Strains were transformed with pVW31, then grown in synthetic glucose minimal medium lacking uracil and containing required amino acids as well as isoleucine and valine (Min D+) (Lucchini et al., 1984). Promega Dual-Luciferase Reporter Assay System Kit (Promega, Madison, WI, USA) was used with a Victor X3 or Victor NIVO Multimode microplate reader (PerkinElmer, Waltham, MA, USA).
Total protein synthesis was measured in wild type and gcn4Δ yeast using a non-radioactive metabolic labeling assay, Click-iT HPG Alexa Fluor 488 Protein Synthesis Assay Kit (Thermo Fisher Scientific) (Mittal et al., 2017). Specifically, this assay utilizes a methionine analog, L-HPG, containing an Alexa Fluor 488 azide. Protein synthesis is determined based on the amount of HPG-Alexa Fluor 488 incorporated and the mean fluorescent intensity was measured by flow cytometry using a BD AccuriTM C6 Flow Cytometer (BD Biosciences) with the support of the UNM Flow Cytometry Shared Resource Facility. Samples were prepared by growing yeast overnight in YPD (yeast extract peptone dextrose) media until mid-log phase at 30C in a New Brunswick model G25-KC Refrigerated/Illuminated Incubator Shaker at 30°C and 300RPM. Yeast were then washed and re-suspended in synthetic media containing 2% glucose and lacking methionine, and were treated with the indicated concentration of borrelidin for 4 h shaking in a New Brunswick model G25-KC Refrigerated/Illuminated Incubator Shaker at 30°C and 300RPM. After 4 h, samples were treated and processed exactly according to the protocol provided by the Click-iT HPG Alexa Fluor 488 Protein Synthesis Assay Kit (Thermo Fisher Scientific) (Mittal et al., 2017). For this assay, only BY4742 MATα background yeast were used as as the corresponding BY4741 MATa strains are auxotrophic for methionine (met15) and are unable to grow in the required methionine-free media. Assays were conducted in triplicate with each sample reflecting the mean fluorescent intensity of 100,000 events.
C. elegans eggs were added to standard S-Complete buffer with 6 mg/mL of UV-killed E. coli strain OP-50 (available from the Caenorhabditis Genetics Center https://cgc.umn.edu/), and grown at 20°C, with observations taken every 24 h. For solid media experiments, C. elegans eggs were added to standard C. elegans NGM media (Brenner, 1974) in 6cm plates (T3306, Tritech Research, Los Angeles, CA, USA) with UV-killed E. coli OP-50, and staged every 24h until adulthood.
Lifespan analysis was conducted at 20°C unless otherwise stated (Hsin and Kenyon, 1999). Specifically, all lifespans were done on 6cm plates (T3306, Tritech Research, Los Angeles, CA, USA) containing standard NGM media (Brenner, 1974) using on-plate UV-killed bacterial food (E. coli strain OP-50 unless otherwise specified) that was first plated at 50μL per lifespan plate in the center of each plate, allowed to grow for 72 h at 25°C, and then treated to 3 × 9999μm Joules ×100 using a UV Stratalinker 2400 (Stratagene, San Diego, CA) on uncovered plates. In all lifespans 0.1mM FUDR (5-fluorodeoxyuridine), (518265 , Bio-World, Dublin, Ohio) was added to plates at Day 1 of adulthood in order to suppress development of progeny. All drug treatments were added to plates after UV treatment and before addition of eggs. Each day, all worms were scored for movement using a 6.7X-45X Trinocular Zoom Stereo Microscope with attached incandescent reflected transillumination base (ZM-2T-EB, AmScope, Irvine, CA), and any dead worms removed using a pick made from a 5.75 inch plain soda-lime glass pasteur pipette handle (Thermo-Fisher) and a platinum wire tip (Tritech Research, Los Angeles, CA, USA). Bagged, missing (crawled off), or exploded worms were noted and censored daily.
All survival curves were plotted using Kaplan-Meier survival curves (Kaplan and Meier, 1958). Statistical significance was determined by calculating p-values using the Wilcoxon Rank-Sum test (Wilcoxon, 1946). Comparisons for developmental timing used Student’s t-test (Student, 1908). Comparisons for yeast doubling time in liquid media and growth on solid media are all versus DMSO only control using Dunnett’s test (Dunnett, 1955). All fold-induction of Gcn4 relative to control is displayed as fold-change with standard error of mean ratio (Sul Lee and Forthofer, 2005). Any noted statistical significance of differences is reported in the corresponding Figure legend. All analysis was performed using R (R Core Team, 2018). |
|
PMC9647509 | Doudou Tan,Wei Wei,Zhen Han,Xuelian Ren,Cong Yan,Shankang Qi,Xiaohan Song,Y. George Zheng,Jiemin Wong,He Huang | HBO1 catalyzes lysine benzoylation in mammalian cells | 26-10-2022 | Biological sciences,Molecular biology,Cell biology | Summary Lysine benzoylation (Kbz) is a newly discovered protein post-translational modification (PTM). This PTM can be stimulated by benzoate and contributes to gene expression. However, its regulatory enzymes and substrate proteins remain largely unknown, hindering further functional studies. Here we identified and validated the lysine acetyltransferase (KAT) HBO1 as a “writer” of Kbz in mammalian cells. In addition, we report the benzoylome in mammalian cells, identifying 1747 Kbz sites; among them at least 77 are the HBO1-targeted Kbz substrates. Bioinformatics analysis showed that HBO1-targeted Kbz sites were involved in multiple processes, including chromatin remodeling, transcription regulation, immune regulation, and tumor growth. Our results thus identify the regulatory elements of the Kbz pathway and reveal the non-canonical enzymatic activity and functions of HBO1 in cellular physiology. | HBO1 catalyzes lysine benzoylation in mammalian cells
Lysine benzoylation (Kbz) is a newly discovered protein post-translational modification (PTM). This PTM can be stimulated by benzoate and contributes to gene expression. However, its regulatory enzymes and substrate proteins remain largely unknown, hindering further functional studies. Here we identified and validated the lysine acetyltransferase (KAT) HBO1 as a “writer” of Kbz in mammalian cells. In addition, we report the benzoylome in mammalian cells, identifying 1747 Kbz sites; among them at least 77 are the HBO1-targeted Kbz substrates. Bioinformatics analysis showed that HBO1-targeted Kbz sites were involved in multiple processes, including chromatin remodeling, transcription regulation, immune regulation, and tumor growth. Our results thus identify the regulatory elements of the Kbz pathway and reveal the non-canonical enzymatic activity and functions of HBO1 in cellular physiology.
Post-translational modifications (PTMs) play important roles in diverse cellular processes, such as transcription, cell cycle, metabolism, and signal transduction (Dancy and Cole, 2015; Lundby et al., 2012; Sabari et al., 2017; Wang and Lin, 2021; Wang and Cole, 2020). Diverse lines of evidence suggest that aberrant PTMs contribute to many diseases (Husmann and Gozani, 2019; Lundby et al., 2019), and their regulatory enzymes, e.g., those for phosphorylation and lysine acetylation (Kac), represent an important class of protein targets for therapeutic drugs (Ferguson and Gray, 2018; Li and Ge, 2020). As has been demonstrated in many well-studied PTM pathways, knowledge of regulatory enzymes and substrate proteins is fundamental to the biochemical characterization of newly discovered PTMs, and offers a stepping stone to revealing their roles in physiology and pathology. Recently, we discovered Kbz as a new type of physiologically relevant PTMs in mammalian cells and identified 22 histone Kbz marks (Huang et al., 2018c). We demonstrated that sodium benzoate (SB), a widely used food additive and a drug approved by the US Food and Drug Administration (FDA) for the treatment of hyperammonemia, could be converted to benzoyl-CoA in mammalian cells and served as the precursor of Kbz (Huang et al., 2018c). ChIP-seq and RNA-seq experiments showed that histone Kbz epigenetic marks are specifically located in promoter regions and are associated with gene expression (Huang et al., 2018c). However, the transferase enzymes that can catalyze Kbz in mammalian cells remain unknown. Moreover, non-histone Kbz substrates are likely to present in mammalian cells but their identities are not known. These knowledge gaps hinder the further functional characterization of this PTM pathway. In this work, we revealed that HBO1, a KAT whose homolog does not exist in S. cerevisiae, acts as a “writer” of Kbz in mammalian cells. In addition, we report the global profiling of benzoylome in mammalian cells, identifying 1747 Kbz sites. Importantly, 77 of these Kbz sites are regulated by HBO1. This study thus discovered both the “writers” and protein substrates of Kbz in mammalian cells, significantly expanding our understanding of Kbz-regulated cellular cascades.
KATs are a group of enzymes that can catalyze Kac reactions on both histone and non-histone proteins (Huang et al., 2018a; Menzies et al., 2016). Emerging lines of evidence demonstrated that some KATs can catalyze multiple types of Kac-independent lysine acylations, such as crotonylation, β-hydroxybutyrylation, lactylation, and isobutyrylation (Xiao et al., 2021; Xie et al., 2016; Zhang et al., 2019; Zhu et al., 2021). Therefore, we hypothesized that some KATs may have benzoyltransferase activity as well. To test this hypothesis, we took advantage of an in vitro fluorometric KATs-catalyzed acylation assay wherein benzoyl-CoA and synthetic human histone H3 or H4 peptides are used as co-factor and substrate, respectively (Gao et al., 2013b). Using this assay, we screened eight KATs, including MOF, Tip60, MOZ, MORF, HBO1, GCN5, PCAF, and HAT1. As expected, all the KATs showed good acetyltransferase activities. On the other hand, only HBO1 and HAT1 showed significant catalytic activities for Kbz (Figure 1A). We next sought to quantitatively compare the acyltransferase activities of HAT1 and HBO1 in the Kac and Kbz reactions. The kinetic analysis results showed that the Kcat values of HAT1 and HBO1 in the Kbz reaction were close to those for Kac. However, the Kcat/Km values of HBO1 and HAT1 were 339s−1M−1 and 407s−1M−1 for Kbz, respectively, which were 9 and 6% of their Kcat/Km values for Kac (Figure 1B). These biochemical results indicated that HBO1 and HAT1 could catalyze Kbz in vitro, although their catalytic activities for Kbz were not as good as for Kac.
To evaluate the cellular benzoyltransferase activity of HBO1 and HAT1, we performed immunofluorescence (IF) staining of Kbz in response to the overexpression of diverse KATs in HeLa cells. Given the broad acyltransferase activity of p300 and CBP (Dancy and Cole, 2015; Huang et al., 2018b), we also include them in the analysis. As expected, ectopic expression of these KATs led to a substantial increase in Kac. Notably, the Kbz levels increased significantly in the cells overexpressing HBO1, p300, and CBP (Figure 2A). However, no obvious change in the Kbz levels could be detected in response to the overexpression of HAT1, GCN5, MOF, PCAF, and Tip60 (Figures 2A and S1). To confirm the Kbz transferase activity of HBO1, CBP, and p300 in cells, we overexpressed them in 293T cells and determined the Kbz levels by western blot (WB). Consistent with the IF results, WB analysis showed that overexpression of HBO1, p300, and CBP, but not HAT1, led to increased histone Kbz levels (Figure 2B). It is not surprising that p300 and CBP catalyzed Kbz in cells because their catalytic pockets are large enough to accommodate diverse acyl-CoAs (Liu et al., 2008). Taken together, our results demonstrate that HBO1 can catalyze Kbz both in vitro and in mammalian cells.
To validate the benzoyltransferase activity of HBO1, we first carried out the Kbz reactions in vitro using core histones extracted from 293T cells and benzoyl-CoA as substrate and co-factor, respectively. Acetyl-CoA was used as a positive control for the assay. WB results showed that wild-type (WT) HBO1 could increase both the Kbz and Kac levels in the core histones (Figure 3A). In addition, its transferase activity was enhanced with the presence of the scaffold protein JADE-1. In contrast, the enzyme-dead mutation (MUT) of HBO1 (G485A/E508Q) abolished its Kbz and Kac transferase activities (Figure 3A). Consistent with the in vitro assay results, when WT HBO1 was overexpressed alone or in combination with JADE-1 in 293T cells, the Kbz levels were increased (Figure 3B). However, no obvious changes in the Kbz level could be detected when the MUT HBO1 (G485A/E508Q) was overexpressed (Figure 3B). Next, to investigate whether HBO1 regulates non-histone Kbz in mammalian cells, we overexpressed WT HBO1 in 293T cells and detected the changes in global Kbz levels. The results showed that overexpression of WT HBO1 elevated the Kbz levels of both non-histone and histone proteins (Figure 3C). All the evidence indicates that Kbz transferase activity is indeed indigenous to HBO1, and it mediates Kbz globally. In addition, molecular docking predicted that the benzoyl-CoA could interact with Thr477, Ile475, Leu511, and Ser512 residues of HBO1 through hydrogen bond interactions, which is similar to acetyl-CoA (Figure 4A). Moreover, a large hydrophobic pocket consisting of Val472, Pro507, Pro510, Leu511, and Gly515 may further stabilize the benzoyl moiety by hydrophobic interactions (Figures 4B and 4C).
Given that HBO1 mediates Kbz globally in mammalian cells, we next asked which Kbz protein substrate and cellular processes are targeted by HBO1. To this end, we performed a quantitative proteomics analysis with three biological replicates to identify the global Kbz sites and quantify their dynamics in response to HBO1 overexpression in 293T cells. In total, 1747 Kbz sites were identified and 1344 of them were quantified (Table S1). The changes of each Kbz site were normalized by the dynamics of corresponding protein levels. Among the quantified Kbz sites, 77 of them were significantly upregulated (log2(overexpression/control) > 1 and p< 0.05) and served as the HBO1-targeted Kbz substrates (Figure 5A). Of interest, H4K8bz and H3K14bz increased by 292.8 and 63.9-folds, respectively (Figure 5B), suggesting that HBO1 may exert epigenetic functions through the regulation of H4K8bz and H3K14bz. These results are reasonable because H4K8 and H3K14 are known major target substrate sites of HBO1 (Tao et al., 2017). Recently, it was reported that the GCN5-containing complex can serve as a histone benzoyltransferase in S. cerevisiae (Wang et al., 2022). However, our results showed that its homolog GCN5 could not catalyze the Kbz reaction in mammalian cells. Moreover, the HBO1 homolog does not exist in yeast. Therefore, we hypothesized that the flanking sequence motifs of Kbz in mammalian cells and yeast are quite different. In support of this notion, the positively charged amino acids Lys and Arg were over-represented and the hydrophobic amino acid Leu was largely depleted in mammalian cells, whereas the hydrophobic and negatively charged amino acids Leu, Ile, Asp, and Glu were enriched in S. cerevisiae (Figure 5C). Gene set enrichment analysis (GSEA) of the Gene Ontology Molecular Function (GO-MF) term indicated that many non-histone Kbz proteins were involved in transcriptional regulation, such as Chromatin remodeling (adjusted p = 2.55E-02) and Regulation of gene expression (adjusted p = 1.80E-03) (Figure 5D). GO Biological Processes (BP) term enrichment analysis of the proteins bearing HBO1-targeted Kbz sites suggested that HBO1 may play important roles in Megakaryocyte differentiation (adjusted p = 2.59E-27), Chromatin remodeling at centromere (adjusted p = 1.76E-26), and Nucleosome assembly (adjusted p = 1.35E-25) through the Kbz pathway (Figure 5E). In addition, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis suggested that HBO1-targeted Kbz proteins were associated with diverse physiological functions, such as Alcoholism (adjusted p = 1.29E-16), Spliceosome (adjusted p = 6.75E-03), and Fatty acid degradation (adjusted p = 4.37E-02) (Figure 5F). These results suggest that HBO1 may regulate diverse cellular physiological and pathological processes by mediating the global benzoylome. Kbz may also affect the cell fate by regulating the functions of key proteins. For example, chromodomain helicase DNA-binding 4 (CHD4), which contains 5 Kbz sites, was confirmed to play a crucial role in chromatin remodeling and regulation of gene expression (Weiss et al., 2020). In addition, Kbz may be associated with diseases such as cancer. B cell lymphoma-2-associated transcription factor 1 (BCLAF1), a death-promoting transcriptional repressor, participates in various biological processes, including autophagy and DNA damage response, and is highly associated with the proliferation and drug-resistance of cancer (Mou et al., 2020; Yu et al., 2022). Notably, it contains eight Kbz sites, therefore linking Kbz to tumor development. Given that some benzoylated lysine residues could also be acetylated, next we systematically compared the benzoylome with known Kac sites based on the UniProt database. The results showed that only 268 Kbz sites could also be acetylated (Table S1), and the proteins with the overlapped sites were mainly involved in chromatin remodeling, gene expression regulation, and metabolism. Of interest, the proteins that only contained Kbz sites were associated with multiple human diseases, such as tumor progression, suggesting that Kbz has unique biological functions from Kac. Emerging evidence indicates that different lysine acylations demonstrate unique cellular functions, even though they have similar structures. Compared with the Kac, Kbz has a larger and more hydrophobic moiety, and therefore, may differentially exert cellular functions based on its structural specificity (Ren et al., 2021). Although both Kac and Kbz are regulated by HBO1, acetyl-CoA and benzoyl-CoA may compete for the active site of HBO1 in the same cellular microenvironment, thus affecting the substrate selectivity of HBO1 and leading to different functional outcomes.
In this study, we identified HBO1 as a Kbz transferase both in vitro and in mammalian cells. Moreover, we report a global benzoylome in mammalian cells, identifying 1747 Kbz sites and revealing at least 77 HBO1-targeted Kbz sites. Functional analysis showed that the HBO1-targeted Kbz sites were associated with diverse cellular physiological and pathological processes, such as chromatin remodeling, transcription regulation, metabolism, immune regulation, tumor progression, and so on. Excessive intake of SB can raise Kbz levels and increase the risk of some diseases, such as motor coordination impairment and ADHD symptoms. Given that HBO1 serves as a Kbz transferase in mammalian cells, our findings thus provide a potential strategy for treating these diseases through the regulation of the Kbz by interfering with HBO1.
In this study, we identified lysine acetyltransferase HBO1 as the “writer” of Kbz in mammalian cells. In addition, we report 1747 Kbz sites in mammalian cells, and at least 77 are HBO1-targeted Kbz substrates. Although bioinformatics analysis revealed that HBO1-targeted Kbz sites are involved in a variety of cellular physiological and pathological processes, these functional sites still require extensive validation at the cellular and animal levels. Further, the roles of some Kbz sites in important proteins remain to be investigated, which may provide some insights for the treatment of certain diseases.
This article does not report additional websites or resources.
Further information and requests for resources and reagents should be direct to and will be fulfilled by the Lead Contact, He Huang ([email protected]).
The plasmids used in this study are listed in the key resources table and are available from the lead contact on request.
HepG2, Hela, and 293T cell lines were purchased from the National Collection of Authenticated Cell Cultures (NCACC) and used without further authentication. Cells were cultured in high-glucose DMEM supplemented with 10% FBS and incubated at 37°C under 5% CO2. No mycoplasma contamination was detected using the reported method (Dreolini et al., 2020).
Unless otherwise noted, all chemical reagents were purchased from Sangon Biotech (Shanghai) Co., Ltd. Antibodies were the following: anti-Kac (1:1000, PTM Biolabs, PTM-101), anti-H3 (1:10,000, Huabio, M1306-4), anti-Actin (1:10,000, Proteintech, 66009-1-Ig), anti-Flag (1:10,000; Sigma-Aldrich, F3165), anti-Myc (1:10,000; Cell Signaling Technology, 2276), anti-Kbz antibody (1:1000, PTM Biolabs, PTM-761), pan anti-Kbz beads (PTM Biolabs, PTM-763), and anti-Flag M2 Affinity Gel (Sigma-Aldrich, A2220).
The benzoylation activity of the KATs was first screened with a fluorogenic assay (Gao et al., 2013a). Synthetic H3-20 or H4-20 peptides containing 20 amino acid residues from the N-terminal of histone H3 and H4 were used as the acyl acceptors. The sequences of H3-20 and H4-20 are Ac-ARTKQTARKSTGGKAPRKQL and Ac-SGRGKGGKGLGKGGAKRHRK, respectively. In the assay, 30 mL of reaction mixture containing 18 mM benzoyl-CoA (Sigma-Aldrich, B1638) or acetyl-CoA (Sigma-Aldrich, A2181), 100 mM peptide substrates, and 100 nM enzymes in the reaction buffer containing 50 mM HEPES, pH 8.0 and 0.1 mM EDTA was incubated at 30°C for 1 h. After the incubation, 30 μL of dimethyl sulfoxide (DMSO) solution containing 50 mM 7-diethylamino-3-(4′-maleimidylphenyl)-4-methylcoumarin (CPM) were mixed with the reaction mixture, followed by co-incubation in darkness at room temperature for 20 min. Addition of DMSO solution quenched the enzymatic reaction and the CPM reacted with the by-product Coenzyme A (CoA-SH) to produce the fluorescent CoA-CPM complex. The fluorescence intensity was then measured with a microplate reader (FlexStation 3) with the excitation and emission wavelength fixed at 392 and 482 nm. The negative control samples were treated in the same way except that the reaction buffer was added to substitute KAT enzymes. Duplicate experiments were performed and the results were summarized in Figure 1A. The benzoylation activity of HBO1 and HAT1 was further characterized with a kinetic assay used in a previous study (Han et al., 2017). In the assay, 200 mM of H4-20 peptide and 100 nM of HAT1 or HBO1 were co-incubated with acetyl-CoA or benzoyl-CoA at varying concentrations for 30 min at 30°C in the reaction buffer containing 50 mM HEPES, pH 8.0 and 0.1 mM of EDTA. 50 mM CPM in DMSO solution was added to quench the enzymatic reaction and to produce the fluorescent CoAS-CPM complex. The fluorescence intensity was measured in the same way as the single-point fluorogenic assay. The kinetic constants Km and kcat were determined with the Michaelis-Menten model and were summarized in Figure 1B.
DNA transient transfection was performed using lipofectamine 2000 (Invitrogen, 11,668,019) according to the manufacturer’s instructions. Immunofluorescence staining and western blot for various proteins were carried out essentially as described (Liu et al., 2013). For immunofluorescence staining, HeLa cells were washed with 1xPBS (137mM NaCl, 2.7mM KCl, 10mM Na2HPO4, and 2mM KH2PO4) before fixation in 4% paraformaldehyde at room temperature for 20 min, incubated with 1% Triton X-100 on ice for 15 min, blocked with 5% BSA in 37°C incubator for 60 min and incubated with the mouse or rabbit anti-Flag/Myc antibody for 2h. The coverslips were washed 3 times with PBST, followed by incubation with Texas Green conjugated secondary antibody against mouse or rabbit. Images were acquired with an Olympus microscope system.
Histones were purified from cells using a standard acid extraction protocol (Shechter et al., 2007). The protein extract (20 μg whole cell protein or 4 μg histone) was fractionated by SDS-PAGE electrophoresis and transferred to the PVDF membrane (GE) using a transfer device according to the manufacturer’s protocol (Biotanon, VE-186). After incubating with 3% BSA in TBST (10 mM Tris, pH 8.0, 150 mM NaCl, 0.5% Tween 20) for 1 h, incubate the membrane with the designated primary antibody (the concentration is shown in the “Reagents” section) overnight at 4°C. Then the membrane was washed 3 times with TBST (5 min each time), and horseradish peroxidase-conjugated anti-mouse or anti-rabbit antibody (1:20,000, Jackson, 115-035-146/111-035-144) was incubated for 1 h at room temperature. Next, the membrane was washed 3 times with TBST (5 min each time) and developed using a chemiluminescence detection system (Biotanon, 4600) according to the manufacturer’s protocol.
Plasmids encoding wild-type or mutant Flag-HBO1 were transfected into 293T cells with or without the plasmid encoding Flag-JADE-1. Cells were collected at 48 h after transfection. Then, the cells were washed with precooled PBS and lysed with lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM sodium chloride, Triton X-100 1%, 1 mM EDTA, 1 mM DTT, 8% glycerol plus protease inhibitor) on ice for 30 min. After centrifugation at 13,400gat 4°C for 10 min, the supernatant was collected and incubated with 10 μL of Flag-M2 beads at 4°C for 2 h. After incubation, the supernatant was discarded and the Flag-M2 beads are washed 3 times with washing buffer (20 mM Tris-HCl pH7.5, 150 mM NaCl, 0.1% Triton X-100, 1 mM EDTA, 1 mM DTT, 8% glycerol plus protease inhibitor). Next, the target protein was eluted by elution buffer (20 mM Tris-HCl pH7.5, 150 mM NaCl, 0.1% NP-40, 1 mM DTT, 10% glycerin plus protease inhibitors).
Histones were extracted from 293T cells using a standard acid extraction protocol (Shechter et al., 2007). The reaction mixtures (including 100 μM of benzoyl-CoA or acetyl-CoA, 2 μg of enzymes as indicated, 100 nM TSA, and 4 μg extracted histone) were incubated in reaction buffer (25 mM Tris-HCl pH 8.0, 150 mM NaCl, 10% glycerol, 1 mM DTT) at 37°C for 1 h. After the incubation, 5×SDS loading buffer was added to the mixture to quench the reaction, and the levels of Kac and Kbz were determined by western blot.
The peptide samples in NH4HCO3 solution were incubated with 30 μL of pan anti-Kbz beads at 4°C overnight. After incubation, the beads were washed three times with NETN buffer (50mM Tris pH 8.0, 100mM NaCl, 1mM EDTA, 0.5% NP40), twice with ETN buffer (50mM Tris pH 8.0, 100mM NaCl, 1mM EDTA), and once with water. The bound peptides were eluted from the beads with 0.1% trifluoroacetic acid and vacuum-dried.
The sample analysis was carried out on an EASY-nLC 1200 UHPLC system (ThermoFisher Scientific) coupled with a Q Exactive HF-X mass spectrometer (ThermoFisher Scientific). Peptides were dissolved in 2.5 μL of solvent A (0.1% FA in water, v/v) and injected into a homemade packed capillary C18 column (20 cm length×75 μm ID, 1.9 μm particle size, Dr. Maisch GmbH, Germany). The quantitative proteome and immunoprecipitated Kbz samples were run in 180- and 120-min gradient, respectively, from 6 to 90% solvent B (A, 0.1% formic acid; B, 80% acetonitrile in 0.1% formic acid). Full mass scans were acquired with 350–1200 m/z at a mass resolution of 60,000. Ions with 2+, 3+, and 4+ charge were selected for MS/MS analysis. The 12 most intensive ions were fragmented with 28% normalized collision energy and tandem mass spectra were acquired with a mass resolution of 15,000. Dynamic exclusion was set to 30 s. The AGC numbers were 3 × 106 and 2 × 105 for MS1 and MS2, respectively. The isolation window was set to 1.3 m/z.
After LC-MS/MS acquisition, the raw files were qualitatively analyzed by MaxQuant software (version1.6.15.0) against the UniProt human database (20,376 entries). Parameters set for quantitative proteomics identification include Trypsin/P as the digestive enzyme; maximum missing cleavage of 2; minimum peptide length of 7; maximum FDR for peptides and proteins of 1%. Cysteine carbamidomethylation was established as a fixed modification. Methionine oxidation and acetylation of the N-terminus were established as variable modifications. Parameter setting of Kbz samples included lysine benzoylation as a variable modification. Other parameters were consistent with the proteome search. FDR thresholds for modification sites were specified at 1%. All the Kbz site ratios were normalized by the quantified protein expression levels.
HBO1 protein was extracted from the crystal structure of human HBO1 in complex with acetyl-CoA (PDB: 5GK9), and the benzoyl-CoA was generated based on the structure of acetyl-CoA using the PyMol package (http://www.pymol.org/). The docking files were prepared with AutoDockTools-1.5.6. The receptor grid file was generated with a box size of 126 × 100 × 100 to cover the protein and calculated by the Autogrid program. The genetic algorithm was selected as “search parameters” (the number of GA runs was set as 100 and other parameters were kept unchanged) and the “docking parameters” was set as default. Molecular docking was performed using the AutoDock program (Morris et al., 2009). The results were evaluated by clustering analysis and visual inspection.
The quantitative proteomics experiments were performed with three biological replicates. All the Kbz site ratios were normalized by corresponding quantified protein expression levels. The quantitative Kbz proteome was analyzed by a two-tailed Student’s ttest for the two groups with HBO1 or vector plasmids transfection. GO, KEGG, and GSEA analyses were adopted with a hypergeometric test in the R clusterProfiler package (Yu et al., 2012). The consensus sequence logo analysis was performed using iceLogo (v1.2) (Colaert et al., 2009).
Experimental values are presented as mean ±SEM. The quantitative benzoylome data were analyzed by a two-tailed Student’s ttest. Differential expression was considered to be significant when p< 0.05, ∗p value less than 0.05; ∗∗p value less than 0.01; ∗∗∗p value less than 0.001. |
|
PMC9647518 | Wen Yuan,Cong-Cong Cui,Jing Li,Yan-Hua Xu,Chun-E Fan,Yu-Chen Chen,Hong-Wei Fan,Bing-Xue Hu,Mei-Yun Shi,Zhi-Yuan Sun,Pei Wang,Teng-Xiang Ma,Zhao Zhang,Min-Sheng Zhu,Hua-Qun Chen | Intracellular TMEM16A is necessary for myogenesis of skeletal muscle | 08-11-2022 | Molecular biology,Cell biology | Summary Transmembrane protein 16A (TMEM16A) localizes at plasma membrane and controls chloride influx in various type of cells. We here showed an intracellular localization pattern of TMEM16A molecules. In myoblasts, TMEM16A was primarily localized to the cytosolic compartment and partially co-localized with intracellular organelles. The global deletion of TMEM16A led to severe skeletal muscle developmental defect. In vitro observation showed that the proliferation of Tmem16a−/− myoblasts was significantly promoted along with activated ERK1/2 and Cyclin D expression; the myogenic differentiation was impaired accompanied by the enhanced caspase 12/3 activation, implying enhanced endoplasmic reticulum (ER) stress. Interestingly, the bradykinin-induced Ca2+ release from ER calcium store was significantly enhanced after TMEM16A deletion. This suggested a suppressing role of intracellular TMEM16A in ER calcium release whereby regulating the flux of chloride ion across the ER membrane. Our findings reveal a unique location pattern of TMEM16A in undifferentiated myoblasts and its role in myogenesis. | Intracellular TMEM16A is necessary for myogenesis of skeletal muscle
Transmembrane protein 16A (TMEM16A) localizes at plasma membrane and controls chloride influx in various type of cells. We here showed an intracellular localization pattern of TMEM16A molecules. In myoblasts, TMEM16A was primarily localized to the cytosolic compartment and partially co-localized with intracellular organelles. The global deletion of TMEM16A led to severe skeletal muscle developmental defect. In vitro observation showed that the proliferation of Tmem16a−/− myoblasts was significantly promoted along with activated ERK1/2 and Cyclin D expression; the myogenic differentiation was impaired accompanied by the enhanced caspase 12/3 activation, implying enhanced endoplasmic reticulum (ER) stress. Interestingly, the bradykinin-induced Ca2+ release from ER calcium store was significantly enhanced after TMEM16A deletion. This suggested a suppressing role of intracellular TMEM16A in ER calcium release whereby regulating the flux of chloride ion across the ER membrane. Our findings reveal a unique location pattern of TMEM16A in undifferentiated myoblasts and its role in myogenesis.
TMEM16A (also called anoctamin 1, ANO1) is a member of the classical Ca2+-activated chloride (Cl−) channel (CaCC) family (Caputo et al., 2008; Schroeder et al., 2008; Yang et al., 2008). It is broadly expressed in many types of cells, localized to the plasma membrane, and contributes to a myriad of biological functions (Duvvuri et al., 2012; Gomez-Pinilla et al., 2009; Jin et al., 2013; Kunzelmann et al., 2009; Tian et al., 2011). TMEM16A molecules adopt a homodimer architecture in which the Cl− _selective pores are surrounded. Each subunit of TMEM16A is composed of ten transmembrane segments and binds two Ca2+ ions (Dang et al., 2017; Paulino et al., 2017). Activation of TMEM16A can be triggered by G-protein-coupled receptor signaling pathways that activate phospholipase C for inositol 1,4,5-trisphosphate (IP3) production and/or Ca2+ release from intracellular stores mediated by the binding of IP3 to IP3 receptors (IP3Rs) (Dayal et al., 2019; Schredelseker et al., 2010; Wang et al., 2018). The resultant TMEM16A activation facilitates the passive flow of Cl− across the plasma membrane, whereby regulates diverse functions such as smooth muscle contraction, nociception, neuronal excitability, insulin secretion, cell proliferation, and migration (Duvvuri et al., 2012; Gomez-Pinilla et al., 2009; Jin et al., 2013; Kunzelmann et al., 2009; Tian et al., 2011). In zebrafish, TMEM16A expressed in mature skeletal musculature locates at plasma membrane near the sarcoplasmic reticulum (SR) and plays a potential role in the excitation-contraction coupling process (Dayal et al., 2019). There are reports showing that TMEM16A also localizes at the plasma membrane of the tether part of endoplasmic reticulum (ER)/SR membrane and facilitates TMEM16A activation by the calcium released from ER/SR (Cabrita et al., 2017; Dayal et al., 2019; Jin et al., 2013; Wang et al., 2020). Interestingly, we here found that TMEM16A could alternatively localize at intracellular organelles of skeletal muscle precursor cells (myoblasts). During embryonic and fetal developmental processes, myoblasts differentiate into mononucleotide myocytes, and then fuse to form multinucleated myotubes and myofibers (Chal and Pourquié, 2017; Kablar and Rudnicki, 2000; Magli and Perlingeiro, 2017; Tran et al., 2013). After birth, muscle myofibers undergo massive growth, fiber type specialization, hypertrophy, and maturation of musculature. In mouse, skeletal muscle development originates at approximately embryonic day 10.5 (E10.5) and is completed at adult (Chal and Pourquié, 2017). Within this process, several transcription factors including muscle regulatory factors (MRFs) with b-HLH domains (e.g., myogenin), and ERK1/2 signaling modules, cyclin D1, caspases, and microRNAs are involved in the regulation of myogenesis (Adi et al., 2002; Holstein et al., 2020; Kablar and Rudnicki, 2000; Li et al., 2012; Skapek et al., 1995; Zhang et al., 1999). As a second messenger, calcium released from ER takes an essential role in myogenesis (Nakanishi et al., 2005, 2007, 2015). ER is the main intracellular calcium store and heavily regulates calcium metabolism through IP3R and RyR systems. IP3Rs can directly mediate Ca2+ release from the ER lumen upon IP3 binding (Bare et al., 2005; Wu et al., 2006), and the released Ca2+ then regulates several biological functions including proliferation and differentiation. However, when the ER calcium homeostasis is impaired, ER stress and apoptosis will be induced (Bare et al., 2005; Santulli et al., 2017; Wu et al., 2006). In addition, the released Ca2+ in cytosol also activates Ca2+-related signaling pathways such as extracellular signal-regulated kinase 1/2 (ERK1/2), thereby interfering the process of cellular proliferation. We here found that intracellular TMEM16A could essentially regulate the IP3R-mediated calcium release from ER and thereby functions in myogenesis. The deletion of TMEM16A led to overproduction of cytosolic calcium induced by augmented release of Ca2+ from ER through IP3R calcium channels. As a consequence, the myoblast displayed enhanced proliferation and inhibited differentiation capacities. Consistently, the TMEM16A knockout mice (KO) exhibited impaired skeletal muscle development. Our findings reveal a novel expression and localization pattern of TMEM16A and its biological relevance in skeletal muscle.
To examine the expression pattern of TMEM16A in skeletal muscle, we firstly measured TMEM16A protein and mRNA of adult skeletal muscles. The results showed that both TMEM16A mRNA and protein were abundantly expressed in the muscles including tongue muscle, tibialis anterior muscle, quadriceps muscle, and soleus muscle (Figures 1A–1C). It was noted that both mRNA and protein levels of TMEM16A in the tongue muscles were much higher than those in other muscles. We next measured TMEM16A in embryonic muscles. The limb muscles at embryonic day (E) 11.5–17.5 expressed TMEM16A protein abundantly, but the expression level decreased apparently by postnatal day 1 (P1) (Figure 1D). To test if this expression pattern paralleled to muscular differentiation process, we used an in vitro model of muscle differentiation by culturing C2C12 myoblast with differentiation medium (DM) and then measured the dynamic expression of TMEM16A. When the myoblasts were cultured with DM, TMEM16A protein level increased immediately and peaked around the third day after the switch. The expression level was reduced afterward (Figure 1E). This observation suggested that the dynamic expression of TMEM16A was associated with skeletal muscle development. We next determined the spatial expression of TMEM16A in skeletal muscle. To our surprise, the immunofluorescence staining of the transverse cross sections showed that, in the immature soleus muscle of 3-week-old mice, most TMEM16A protein was expressed randomly in the cytosol of myofibers rather than the sarcolemma (Figure 1F). In the 8-week-old mice, however, TMEM16A localized exclusively in the sarcolemma of mature soleus myofibers in a similar manner of laminin. To further resolve the TMEM16A expression within myofibers, we isolated individual myofibers from the extensor digitorum longus (EDL) muscle of 8-week-old mice and subjected to immunofluorescence staining. TMEM16A located at the sarcolemma of mature sarcomeres in a very good order, and it did not overlap with either DHPRα1 (dihydropyridine receptor, T-tubule marker protein) or RyRs (ryanodine receptors, SR marker protein) (Figures 1G and 1H). Thus, we concluded that TMEM16A did not reside in the T-tubule or SR of mature myofibers. Results from co-staining of the myofibers with anti-TMEM16A and actinin antibodies consistently indicated that TMEM16A did not locate to the Z-discs corresponding regions of the sarcolemma (Figure S1). In immature myofibers isolated from 3-week-old mice, part of TMEM16A molecules showed similar location as mature myofibers. However, the other TMEM16A molecules were diffusely distributed in the cytosol of cells. Combined with the results from the cross-section staining, our observations indicated an intracellular localization of TMEM16A in the developing skeletal muscle cells. We here did not study the biological relevance of TMEM16A localization in mature myofibrils.
The intracellular localization of TMEM16A in immature myofibers allowed us to predict that TMEM16A might exist in subcellular organelles. As there was evidence showing that TMEM16A had capacity to bind IP3R (Cabrita et al., 2017), we guess that TMEM16A may locate at ER or SR of myoblasts. To validate this point, we firstly transfected the vector-expressing GFP fusing with full length of TMEM16A (GFP-TMEM16A) to primary cultured myoblasts. GFP signals were detected abundant in the cytosol and modest in the plasma membrane (Figure 2A). This expressing pattern was unlikely attributable to the abnormality of the recombinant fusing protein because typical plasma membrane localization of this protein could be measured in Chinese hamster ovary cells (CHOs) transfected with the same plasmid (Figure S2). When the proliferative myoblasts underwent myogenic differentiation for 4 days (DM4), a large partial of the fluorescence signals moved to the plasma membrane of the newly formed myotubes. Interestingly, the CaCC activity in myoblast was almost undetectable in primary cultured myoblasts but apparent in myotubes (Figure S3), verified the distinguished localization patterns of TMEM16A in the skeletal muscle cells at different developmental stages. To explore the intracellular localization of TMEM16A in myoblast, we co-transfected the primary cultured myoblasts with expressive vectors of recombinant GFP-TMEM16A and mCherry-Sec61β, a subunit of the Sec61 translocation channel on the ER membrane (Gemmer and Förster, 2020). Interestingly, the GFP-TMEM16A molecules were overlapped with mCherry-Sec61β signals (Figure 2B), whereas almost no co-localized signals were detected in myotubes after 4 days of myogenic differentiation induction. We next co-stained the primary cultured myoblasts of mice with anti-TMEM16A and IP3Rs (IP3RI/II/III) antibodies. The results showed that only a part of TMEM16A signals overlapped with those of IP3Rs (Figures 2C and 2D). This result indicated that a part of intracellular TMEM16A might locate at ER and others might locate at other intracellular organelles.
As ER is the primary intracellular calcium store, we speculate a role of TMEM16A in calcium release from ER. To test this hypothesis, we isolated myoblast cells from wild-type control (CTR) and TMEM16A knockout mice (KO) and cultured them in vitro, then stimulated the primary cells with bradykinin, an agonist of IP3Rs Ca2+ channels (Tarroni et al., 1997). To our surprise, upon stimulation with 100 nM bradykinin, the mutant myoblast cells displayed stronger cytosolic calcium signals in contrast to the control (Figure 3A). Quantitation analysis showed that both the signal amplitude and the area under the signal curve of the myoblast cells were significantly increased in the mutant myoblast (Figures 3B and 3C). No apparent intracellular calcium concentration elevations were observed in both the mutant and control myoblasts against stimulation with 20 mM caffeine (a RyR agonist) (Meissner, 2017) (Figures 3D–3F), indicating the undifferentiated state of the myoblasts. It was notable that both the bradykinin and caffeine-induced Ca2+ signals in the mutant myotubes were much less than the controls (Figures 3G–3L). Interestingly, the protein expression level of IP3Rs was slightly declined in the KO myoblast compared with WT control (Figure S4), suggesting that the elevated calcium release capacity of IP3Rs may attribute to the functional alteration of the channel. In addition, the expression of RyRs was also decreased in KO myotubes, consistent with the downregulated calcium response against caffeine. This result suggested that TMEM16A was not involved in the modulation of calcium release through RyR calcium channels. Our above observation clearly indicated that TMEM16A served to inhibit IP3R-mediated release of Ca2+ from intracellular stores of myoblasts. However, no physical interaction was detected between endogenous TMEM16A and IP3Rs molecules by immunoprecipitation (Figure S5).
To determine the role of TMEM16A in skeletal muscle development in vivo, we analyzed the muscular phenotypes of the mice with global deletion of Tmem16a gene. The deletion efficiency in skeletal muscle was confirmed by immunoblotting, real-time quantitative RT-PCR, and immunofluorescence staining (Figures 4A–4C). Accordingly, the CaCC activity was significantly declined in the myotubes-deleted TMEM16A (Figure S6). Consistent with previous observations of another group (Rock et al., 2008), we found that the KO neonates (P1) were seemingly healthy and had comparable body weights to the control littermates (Tmem16a+/+). However, the KO mice were visibly distinguishable from the control littermates with lower body weights as the age increased (Figures 4D and 4E). Moreover, the skeletal muscle mass was also smaller in the KO mice (Figures 4F and 4G). Further examination showed that all the mutant muscles were smaller than the control muscles. Figure 4G represents the gross morphologies of mutant and control tongue muscles and limb muscles of 5-day-old mice. These data together indicated that the reduced muscle mass was a primary cause for the lower body mass in the KO mice. Moreover, histological analysis showed that the limb muscles of neonatal KO mice contained sparse and disorganized myofibers embedded within an amorphous mass of extracellular matrix. Similarly, the tongue muscle of the KO mice was smaller than the control, and the myofibers were shorter and disorganized (Figure 4H). We also examined the skeleton of P10 mice and observed apparent abnormality in the mutant mice including the shorter bones and kyphosis of cervical vertebrae (Figure 4I). Since TMEM16A is not expressed in the chondrogenic mesenchyme at any time during mouse development (Rock et al., 2008), this abnormality was likely secondary to the defect of skeletal muscle. Given the body weight of KO mice was comparable to the WT mice at perinatal stage, we also performed the histological analysis of the muscles. The results showed that the gross morphology of the KO limb muscles was almost normal. However, the muscles were smaller, and the sparse and disorganized myofibers embedded within an amorphous mass of enlarged extracellular matrix were also observed in the KO muscles, similar to the elder mice (Figure S7). Above observations collectively showed that the ablation of TMEM16A caused an apparent defect of skeletal muscle development in mice both during embryonic stage and postnatal growth.
The developmental defect of the Tmem16a−/− skeletal muscle implies impaired proliferation and/or differentiation of muscular precursors. As Ca2+ release from intracellular stores has diverse roles in regulating tremendous physiological processes including cell proliferation and differentiation (Nakanishi et al., 2005, 2007, 2015; Wu et al., 2006), we firstly assessed the effect of TMEM16A on primary cultured myoblasts expansion with 5-ethynyl-2′-deoxyuridine (EdU) incorporation assay. The result showed that the proliferation capability was apparently enhanced in the mutant myoblasts isolated from KO mice compared with the control cells (Figure 5A). We then examined the expression of phosphorylated ERK1/2 (p-ERK1/2) and CyclinD1, the key signal modules of cell proliferation. The results showed that the levels of p-ERK1/2 and CyclinD1 were significantly upregulated in Tmem16a−/− myoblasts (Figures 5B–5E). These observations indicated that the augmented elevation of cytosolic Ca2+ by TMEM16A deletion led to enhancement of myoblast proliferation. We next investigated the effects of TMEM16A deletion on the myogenic differentiation. The differentiation process was inhibited in the KO myoblast as evidenced by the smaller size of the formed myotubes and the decreased fusion index (Figures 5F and 5G), and reduced myogenin and myosin heavy chain protein (MyHC) (Figures 5H–5K). Proper apoptosis of myoblasts was essential for myogenic differentiation (Hochreiter-Hufford et al., 2013; Ikeda et al., 2009). We then examined the apoptosis-associated molecules during myogenesis progression. As shown in Figures 6A and 6B, the activation of caspase 3 (cleaved caspase 3) was significantly heightened in the myoblasts cultured from the KO mice at DM1 and DM2 compared with control. The impaired myogenic differentiation of the KO myoblasts was significantly improved by the administration of Z-VAD-FMK, a pan-caspase inhibitor (Figures 6C and 6D). Moreover, as ER stress is the main initial event attributed to myogenic differentiation-induced apoptosis that is characterized by the expression and activation of the initiator, capase12, and the downstream effector, caspase 3 (Nakagawa et al., 2000; Nakanishi et al., 2005, 2007), we then measured these proteins in the differentiating myoblasts. As expected, both the expression levels of pro-caspase 12 and the activated caspase 12 (cleaved caspase 12) were significantly elevated at the early stage of myogenesis (DM1) in KO myoblasts, which was in parallel to the elevated caspase 3 activation (Figures 6E and 6F). These observations suggested that the excessive ER stress and apoptosis caused by TMEM16A deletion underlies the impaired differentiation of TMEM16A KO skeletal muscle. Collectively, above findings indicated that Tmem16A deletion altered the pathways of cell fate commitment and ER stress, and the latter might be more prone to cell death secondary to inhibition of differentiation.
In this report, we demonstrated that TMEM16A could localize to the intracellular organelles of undifferentiated myoblast, revealing a novel expression pattern of TMEM16A. Biochemical analysis showed that the deletion of TMEM16A led to an enhanced elevation of cytosolic calcium after agonistic activation of IP3R, the resultant calcium promoted myoblast proliferation, and the excessive ER stress inhibited myoblast differentiation as well. We thus proposed a role of such intracellular TMEM16A was to control ER calcium homeostasis through the IP3R calcium channels-mediated Ca2+ release, and such an elaborative modulation was particularly important for coordination of myogenic processes. These findings revealed an alternative regulatory scheme for skeletal muscle development. According to current knowledge and the evidence provided by this report, a working model for this scheme may be proposed (Figure 7). In this model, partial of intracellular TMEM16A is expressed in the ER of myoblast cells and functions to maintain intracellular calcium homeostasis through facilitating the influx of chloride, thereby coordinating proliferation and differentiation processes of skeletal muscle precursor cells. After these processes have been completed, the intracellular TMEM16A protein amount was reduced and then exclusively expressed at sarcolemma membrane of myofiber, to regulate the process of muscle contraction (Dayal et al., 2019). When the intracellular TMEM16A is inhibited, the influx of chloride to ER is blocked, resulting in over release of ER calcium. The elevated cytosolic calcium then promotes muscular precursors proliferation via upregulation of cell cycle-related signals/factors (e.g., p-ERK1/2 and CyclinD1). Simultaneously, the imbalance of calcium homeostasis in ER caused excessive ER stress showed by overproduction/activation of caspase 12/caspase 3, resulting in inhibition of myogenic differentiation. Among these processes, the intracellular TMEM16A serves as a negative regulator of IP3Rs-mediated calcium release that is necessary for coordinating the processes of proliferation and differentiation. Previously, it was reported that inhibition of caspase 3 inhibited the formation of myotubes (Fernando et al., 2002; Hochreiter-Hufford et al., 2013). We here show that intracellular-resident TMEM16A functions to suppress calcium release from ER lumen thus inhibit the activation of caspase 3. The discrepancy between our study and previous study might be explained by the reasons below. Although the caspase 3 activation is required for the myogenesis progression, the caspase 3 activation level should be maintained at a proper level. Too strong caspase 3 activity led to cell death and inhibition of myogenic differentiation. We did not observe inhibition effects of the pan-caspase inhibitor (Z-VAD-FMK) on the differentiation in control wild-type myoblasts as other labs. The reason may be related to the degree of inhibition effect of lower concentration of the inhibitor than other labs. Megeney lab used caspase-3-specific inhibitor, Z-DEVD.FMK at concentration of 20 μM, while Ravichandran lab used the same agent as ours, Z-VAD-FMK at 50 μM (Fernando et al., 2002; Hochreiter-Hufford et al., 2013), both of which had much stronger inhibitory functions than that of our current study (Z-VAD-FMK at 10 μM). It seems that the caspase 3 functions in a dose-dependent manner. Within this working model, how does the TMEM16A molecule inhibit IP3Rs-mediated calcium release is an interesting issue. In the plasma membrane of neurons, the featured localization of TMEM16A at the proximal region of the ER membrane through physical interaction with IP3R allows TMEM16A to be activated more efficiently by IP3R-mediated calcium (Jin et al., 2013). However, no physical interaction between TMEM16A and IP3Rs was detected in myoblast in this study. We thus proposed the ER-resident TMEM16A to be activated by IP3R-mediated calcium release. The resultant activation of TMEM16A allows chloride to move into ER lumen and bind Ca2+, and hence the ER calcium release is inhibited. As a matter of fact, similar effect was observed in the case of intracellular CFTR (cystic fibrosis transmembrane conductance regulator) (Benedetto et al., 2017; Divangahi et al., 2009; Huang et al., 2020).The intracellular CFTR promotes Ca2+ influx to ER through facilitating the flux of chloride and its binding to Ca2+ in the ER in smooth muscle cells. As CFTR is also expressed in skeletal muscle cells (Divangahi et al., 2009), whether it is involved in the role of TMEM16A in ER calcium release remains to be determined in the future. As a second messenger, cytosolic calcium functions in diverse physiological processes. Undoubtedly, elaborative control and homeostasis of this messenger is extremely important for various physiological activities. In this study, we show that intracellular TMEM16A is involved in the modulation of calcium release from intracellular ER through IP3Rs in a chloride channel activity-dependent manner. It would be interesting to extend this mechanism to other type of cells. In summary, we here revealed a unique expression pattern of TMEM16A in skeletal muscle precursors and defined a role of intracellular TMEM16A in myogenesis. The deletion of TMEM16A caused overproduction of cytosolic calcium and excessive ER stress through abolishment of the chloride channel activity of TMEM16A, thereby enhancing cell growth and inhibiting differentiation.
In this study, we observed a unique localization of TMEM16A molecules in the cytosolic organelles of undifferentiated myoblast. However, we did not delineate the individual organelles targeting TMEM16A by organelle fraction analysis. It would be interesting to address this limitation in the future studies. Another limitation of this study is lacking measurement of chloride concentration in the ER of differentiating myoblast cells due to our shortage of measurement method, which would weaken our proposal that the intracellular TMEM16A modulates the chloride flux across ER membrane.
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Huaqun Chen ([email protected]).
This study did not generate new unique reagents.
The animal experiments were approved by the Experimental Committee of Nanjing Normal University (No: IACYUC-1903023; Approve date: 1 March 2019).
This study used C2C12 myoblast cell line and CHO-K1 hamster ovary epithelial cell line.
3-week and 8-week-old male C57BL/6 (B6) mice were purchased from Gem Pharmatech (Nanjing, China). Tmem16a conventional knockout mice were generated as previously reported (Zhang et al., 2016). The mice were maintained in the specific pathogen free (SPF)-grade animal facility at Nanjing Normal University. All animal procedures were performed in accordance with the guidelines of the Animal Care and Use Committee of Nanjing Normal University. The animal experiments were approved by the Experimental Committee of Nanjing Normal University. Deletion of TMEM16A was evaluated by qRT-PCR, Western blotting, and immunofluorescence staining.
Since the TMEM16A KO mice were neonatal lethal, we performed all the analysis in one month. The body weight mass was measured at 1, 2, 5-, 10-, 20- and 26-days old age. The muscle tissues were dissected and fixed, transverse cross sections or cryosections were prepared, followed by the histological or immunofluorescence staining analysis with specific antibodies. The pups were fixed in 95% ethanol for 8–12 h, the skin and all the tissues were removed. Then the skeletons were fixed in 95% ethanol for 3 days, defat in acetone for 1 day, washed with tap water for 8 h, stained with Alcian blue (Sigma Aldrich #TMS-010C) and Alizarin red (Solarbio #G1450, China) for 6 days, washed in tap water for 30 min, decolorized in 2% potassium hydroxide for 12 h. The skeleton images were captured under a stereoscope (Motic, China).
For preparation of cryosections, the fresh tissues were flash frozen in isopentane soaked in liquid nitrogen, embedded in OCT Compound (Sakura, USA). The samples were sectioned at 11 μm thick by a freezing microtome (Leica CM1950). For paraffin sections, the fresh tissues were fixed with 4% PFA (paraformaldehyde) (Biosharp #BL539A, China) and processed for routine paraffin histology. Paraffin sections (7 μm) were stained with hematoxylin-eosin (Beyotime #C0105S, China) using routine procedures. Immunofluorescence staining was performed by fixation of the cryosections or cells with 4% PFA, permeabilization with 0.2% Triton X-100 in PBS, blocking with 2.5% normal goat serum (Thermo Fisher #R37624), incubation with a primary antibody at 4°C overnight, and incubation with the corresponding Alexa Fluor-conjugated secondary antibodies (Thermo Fisher #A-11008, A-21422; CST #4413, #4408). The slides were then incubated with DAPI (Sigma Aldrich #D9542) to visualize the cell nuclei. The primary antibodies used were specific for TMEM16A (Abcam #ab53212), myosin heavy chain (MyHC) (DSHB #MF20) and Laminin (Sigma Aldrich #L9393). The slides were washed and mounted with ProLong Gold Anti-fade Mountant (Thermo Fisher #P10144). Images were acquired under a confocal microscope (Nikon A1). Image processing was performed by using Adobe Photoshop software.
Whole muscle homogenates were prepared from the soleus, gastrocnemius, tibialis anterior and hind limb muscles. Aliquots taken for the whole muscle homogenates were combined with SDS sample buffer and incubated at 100°C for 5 min, with two rounds of vigorous vortex during the incubation. The homogenates were centrifuged at 12 000 rpm for 20 min, and the supernatants were used as the whole muscle homogenate samples.
The single myofibers were prepared according to previous method (Pasut et al., 2013). Briefly, EDL muscles were digested with an enzyme mixture (0.05% collagenase D (Sigma Aldrich #11088858001), 0.25 μM CaCl2 and 0.5U/ml Dispase (Sigma Aldrich #SCM133)) in high-glucose DMEM (Thermo Fisher #11965092) at 37°C for about 30 min. The released single myofibers were rinsed with DMEM, followed by recovered in the incubator at 37°C for 30 min.
Skeletal muscle tissues were removed from the limbs of neonatal mice. The tissues were cut into small pieces and digested in the enzyme mixture (same as above) at 37°C for 0.5–1 h. 10 mL of DMEM was added to the digested mixture and triturated prior to centrifugation at 1500 × g for 10 min at room temperature. The pellet was resuspended in Ham’s F-10 nutrient mixture (Thermo Fisher #41550088) containing 20% FBS, bFGF (R&D #3339-FB) (5 ng/mL) and 1% penicillin/streptomycin (Thermo Fisher #15140122) (growth medium, GM) and plated in a 10 cm culture dish for 1 h. The unattached cells were harvested and cultured in Matrigel-coated 24-well culture plates. When the cell confluence reached approximately 80%, the cells were switched to high-glucose DMEM containing 2% horse serum (Sangon #ES10006-0100, China) and 1% penicillin/streptomycin (differentiation medium, DM) to induce myogenesis. C2C12 cells were cultured in GM, and myogenic differentiation was induced in DM.
The GFP-Tmem16a recombinant plasmid or mCherry-Sec61β recombinant plasmid was transfected into cultured primary myoblasts using LipoMax transfection reagent (Sudgen # 32011, China) according to the manufacturer’s instructions. Cells were collected for subsequent analysis after 48 h of transfection. The transfected myoblasts were switched to DM for another 4 days to induce myogenic differentiation.
The primary cultured myoblasts (the 2nd passage) were seeded into 24-well plates at 1×105 cells per well and incubated for 24 h. The 5-ethynyl-20-deoxyuridine (EdU) incorporation assay was performed with an EdU assay kit (Beyotime #C0071S, China) according to the manufacturer’s manual. Briefly, EdU (10 μM) was added into the wells and incubated for 2 h, followed by counterstaining of the cell nuclei with Hoechst 33342 (Beyotime #C1025, China). Images were captured under a fluorescence microscope (Leica). The percentage of EdU-positive cells was calculated in five fields per well.
Protein samples from tissues or cells were subjected to electrophoresis on 4%, 8% or 12% (wt/vol) polyacrylamide gels and transferred onto a PVDF membranes (Millipore # IPVH00010, USA). Membranes were blocked with 5% (wt/vol) skim milk and sequentially reacted with specific primary antibodies and the appropriate horseradish peroxidase–conjugated secondary antibody (Santa Cruz #sc-2004, sc-2005). The signals were visualized by incubation in Clarity Western ECL Substrate (Beyotime P0018S, China) prior to scanning (Tanon 4500). The primary antibodies used were specific for TMEM16A (Abcam# ab53212), myosin heavy chain (DSHB #MF20), myogenin (DSHB #F5D), RyR (Santa Cruz #sc-376507), IP3RI/II/III (Santa Cruz #sc-377518), Cyclin D1 (CST #2978), caspase 3/cleaved caspase 3 (CST #9662) and caspase 12 (Santa Cruz #sc-21747). Band intensities were determined by ImageJ software.
To assess mRNA expression of the genes, total RNA from tissues or cells was extracted with TRIzol Reagent (Invitrogen). cDNA was synthesized by using a PrimeScript RT reagent kit (Takara #RR037B), and the transcription was quantified by quantitative PCR using a SYBR green kit (Vazyme #Q221-01, China) in a StepOnePlus thermal cycler (Applied Biosystems, USA) according to the manufacturer’s instructions. The specific PCR primers are listed in Table S1.
Ca2+ signals were measured by using the Ca2+ indicator dye Fluo-4 (Fluo-4 Direct Calcium Assay Kit, Thermo Fisher) according to the manual and previous study (Spoida et al., 2014). Briefly, undifferentiated or differentiated myoblasts (myotubes) grown in a 96-well black plate (clear bottom with lid, Corning) were incubated with 50 μL Fluo-4 calcium indicator dye in a CO2 incubator for 45 min at 37°C. Baseline fluorescence was detected every 2 s for 20 s on a luminescence spectrometer (BioTek Synergy H1 393176 microplate reader) with excitation at 490 nm and emission at 525 nm at room temperature. The agonists were added to the wells and the fluorescence signals were recorded every 2 s for 300 s. For each condition, fluorescence counts were normalized to the basal fluorescence signal. The elevation of intracellular calcium was indicated by the increase in fluorescence counts.
Cells were rinsed twice with ice-cold PBS and protein samples were prepared according to our previous study (Chen et al., 2006). Proteins (500 μg) were immunoprecipitated for 2 h with IP3R antibody (1 μg). The precleared Protein A/G agarose beads (Thermo Fisher #20423) were incubated with immunocomplexes for another 2 h and washed four times with the lysis buffer. The samples were separated by SDS-PAGE, transferred to a PVDF membrane, and detected by Western blot analysis.
Statistical analyses of the data were performed using GraphPad Prism software. Western blotting and Co-IP data are representative of three independent experiments. All biological replicates and group sizes used for statistical analyses are indicated in the figure legends. The results are presented as the mean ± SD values. Student’s t-test (for paired or unpaired samples) was used to determine the significance of differences between two groups; a p value less than 0.05 was considered statistically significant. |
|
PMC9647533 | Meijuan Xi,Ping Zhao,Fang Li,Han Bao,Sijie Ding,Lijiang Ji,Jing Yan | MicroRNA-16 inhibits the TLR4/NF-κB pathway and maintains tight junction integrity in irritable bowel syndrome with diarrhea | 05-09-2022 | microRNA-16,toll-like receptor 4,NF-κB,X-inactive specific transcript,irritable bowel syndrome with diarrhea,AWR, abdominal withdrawal reflex,CCK-8, cell-counting kit-8,cDNA, complementary DNA,EMG, electromyography,FBS, fetal bovine serum,IBS, irritable bowel syndrome,IBS-D, IBS with diarrhea,LPS, lipopolysaccharide,NC, negative control,PDTC, pyrrolidine dithiocarbamate,qRT-PCR, quantitative RT-PCR,TER, transepithelial resistance,XIST, X-inactive specific transcript | Irritable bowel syndrome with diarrhea (IBS-D) is a chronic and relapsing inflammatory disorder in which pathogenesis has been shown to be in part the result of miRNA-mediated signaling. Here, we investigated the alleviatory role of miR-16 in IBS-D. First, we established an IBS-D mouse model using colonic instillation of acetic acid and developed an IBS-D cell model using lipopolysaccharide exposure. The experimental data demonstrated that miR-16 was underexpressed in the serum of IBS-D patients, as well as in the colorectal tissues of IBS-D mouse models and lipopolysaccharide-exposed intestinal epithelial cells. Next, miR-16 and TLR4 were overexpressed or inhibited to characterize their roles in the viability and apoptosis of intestinal epithelial cells, inflammation, and epithelial tight junction. We found that miR-16 overexpression increased the viability of intestinal epithelial cells, maintained tight junction integrity, and inhibited cell apoptosis and inflammation. We showed that miR-16 targeted TLR4 and inhibited the TLR4/NF-κB signaling pathway. Additionally, inhibition of NF-κB suppressed the long noncoding RNA XIST, thereby promoting enterocyte viability, inhibiting apoptosis and cytokine production, and maintaining tight junction integrity. In vivo experiments further verified the alleviatory effect of miR-16 on IBS-D symptoms in mice. Taken together, we conclude that miR-16 downregulates XIST through the TLR4/NF-κB pathway, thereby relieving IBS-D. This study suggests that miR-16 may represent a potential target for therapeutic intervention against IBS-D. | MicroRNA-16 inhibits the TLR4/NF-κB pathway and maintains tight junction integrity in irritable bowel syndrome with diarrhea
Irritable bowel syndrome with diarrhea (IBS-D) is a chronic and relapsing inflammatory disorder in which pathogenesis has been shown to be in part the result of miRNA-mediated signaling. Here, we investigated the alleviatory role of miR-16 in IBS-D. First, we established an IBS-D mouse model using colonic instillation of acetic acid and developed an IBS-D cell model using lipopolysaccharide exposure. The experimental data demonstrated that miR-16 was underexpressed in the serum of IBS-D patients, as well as in the colorectal tissues of IBS-D mouse models and lipopolysaccharide-exposed intestinal epithelial cells. Next, miR-16 and TLR4 were overexpressed or inhibited to characterize their roles in the viability and apoptosis of intestinal epithelial cells, inflammation, and epithelial tight junction. We found that miR-16 overexpression increased the viability of intestinal epithelial cells, maintained tight junction integrity, and inhibited cell apoptosis and inflammation. We showed that miR-16 targeted TLR4 and inhibited the TLR4/NF-κB signaling pathway. Additionally, inhibition of NF-κB suppressed the long noncoding RNA XIST, thereby promoting enterocyte viability, inhibiting apoptosis and cytokine production, and maintaining tight junction integrity. In vivo experiments further verified the alleviatory effect of miR-16 on IBS-D symptoms in mice. Taken together, we conclude that miR-16 downregulates XIST through the TLR4/NF-κB pathway, thereby relieving IBS-D. This study suggests that miR-16 may represent a potential target for therapeutic intervention against IBS-D.
Irritable bowel syndrome (IBS) is a chronic and relapsing bowel disorder affecting 11% of the global population with notable disease burdens, for example, decreased productivity and reduced life quality (1). IBS is characterized by chronic abdominal discomfort and alternations in frequency and appearance of stool (2). Accordingly, IBS is further classified into three subtypes: IBS with diarrhea (IBS-D), IBS with constipation, and IBS with a mixed bowel pattern. Among these subtypes, IBS-D accounts for a quarter to a half of all IBS cases (1, 3). The pathogenesis of IBS-D remains to be fully understood. Multiple etiological factors and triggers have been identified, including genetic susceptibility, visceral hypersensitivity, increased mucosal permeability, and altered gut microbiology (4). Emerging evidence has highlighted the significance of miRNAs as potential biomarkers against IBS (5). Martinez et al. (6) analyzed the differentially expressed miRNAs in IBS-D patients by RNA sequencing and identified miR-125b-5p and miR-16-5p as the most downregulated miRNAs in the context of IBS-D. The aforementioned findings encouraged us to undertake miRNA-involving IBS-D studies. miRNAs refer to short strands of RNA with the length of about 22 nucleotides that function as guide molecules to knock down the target mRNA and downregulate the corresponding proteins (7). Moreover, miRNAs were matured in the cytoplasm and can be transported by extracellular vehicles to the neighboring cells or into the circulatory system (8). Of note, miRNAs in the serum are recognized as potential targets for therapeutic agents for a variety of diseases. In this sense, we started an investigation from microRNA-16 (miR-16) in the serum of IBS-D patients and explored its downstream effectors. In silico prediction of this study identified toll-like receptor 4 (TLR4) as a putative target of miR-16. Interestingly, TLR4 is a pathogen-recognition receptor of inflammation, which is contributory to IBS (9). In addition, intestinal barrier function could be recovered by wogonin through inactivating TLR4-dependent NF-κB pathway (10). Furthermore, NF-κB signaling pathway has been involved in diverse pathological responses, such as cancer and chronic inflammation, and has a role to confer in IBS (11, 12). Moreover, it has been documented that NF-κB signaling activated the expression of X-inactive specific transcript (XIST) (13), and XIST was positively related to the IBS (14). Thus, we hypothesized in the present study that miR-16 may present implications in biological processed of IBS-D, which may involve the TLR4/NF-κB/XIST axis.
In order to reveal the role of miR-16 in IBS, we first measured the miR-16 expression in the serum from IBS-D patients. As shown in Figure 1A, miR-16 was underexpressed in the IBS-D patients. We then developed an IBS-D mouse model. IBS-D mice were depressive with decreased hair gloss and reduced food intake. Defecation intervals, stool types, and abdominal withdrawal reflex (AWR) of the mice were evaluated (Fig. 1, B–D). IBS-D mice presented with shorter defecation intervals, increased water content in stool, and higher AWR scores. Results of the colorectal stepwise distention (Fig. 1E) indicated that the injury threshold of IBS-D was decreased in comparison to the control mice. Electromyography (EMG) activity (Fig. 1F) in the IBS-D mice was enhanced versus the control mice. As illustrated by ELISA and quantitative RT-PCR (qRT-PCR) (Fig. 1G), the cytokine production was elevated in the intestinal tissue homogenate of IBS-D mice, and levels of IL-1β and IL-6 were observed to be upregulated in IBS-D mice relative to the control mice. H&E staining assay was adopted to study the histologic change of the colon mucosa (Fig. 1H). No obvious inflammation or pathological changes were observed. These results indicated that the IBS-D mouse model was successfully established. The level of miR-16 in the mouse colorectal tissue was measured by the qRT-PCR (Fig. 1I), and miR-16 was revealed to be poorly expressed in the IBS-D mice. Moreover, the serum level of miR-16 in IBS-D mice was confirmed to be downregulated as compared with that in control mice (Fig. 1J). Taken together, our data demonstrated the downregulation of miR-16 in both IBS-D patients and mice.
To investigate the effect of miR-16 on the intestinal epithelial cells, we established the lipopolysaccharide (LPS)-induced IBS-D model in the normal colonic epithelial cell line NCM460 (15), where miR-16 was subsequently overexpressed. Relative to the control group, miR-16 expression in the LPS-treated intestinal epithelial cells was remarkably decreased. Relative to the LPS + mimic negative control (NC) group, miR-16 mimic upregulated the miR-16 expression (Fig. 2A). Cell-counting kit-8 (CCK-8) assay and flow cytometry (Fig. 2, B and C) illustrated that cell viability in the LPS-induced intestinal epithelial cells was impeded and the apoptosis was accelerated versus the control cells, which could then be reversed in response to miR-16 mimic transfection. As revealed by ELISA and qRT-PCR (Fig. 2D), expression of IL-1β and IL-6 was enhanced in the LPS-treated intestinal epithelial cells. Whereas, miR-16 mimic suppressed IL-1β and IL-6 levels. Epithelial tight junction proteins ZO-1 and occludin were downregulated in the IBS-D cell model, and their levels were restored after the cells were treated with miR-16 mimic, as verified by Western blot (Fig. 2E). Further, the tight junction integrity was compromised by LPS treatment, as reflected by reduced transepithelial resistance (TER) level, and additional miR-16 mimic transfection led to restored TER level, indicating enhanced tight junction integrity (Fig. 2F). These data supported the finding that overexpression of miR-16 promoted enterocyte viability, inhibited apoptosis and cytokine production, and maintained tight junction integrity.
The downstream modulation factors of miR-16 were predicted by RNAInter (score > 0.8). The interaction of the 161 downstream genes was analyzed by STRING and visualized by Cytoscape 3.7.2 software package (https://cytoscape.org/), as illustrated in Figure 3A. There were 23 genes in the core position of the network (degree > 20). Searching in GeneCards yielded 100 IBS-related genes. Intersection of the 23 core genes and the 100 IBS-related genes revealed eight candidate genes, namely TP53, KRAS, MTOR, TNF, BRCA1, BDNF, PTPN11, and TLR4 (Fig. 3B). To further understand the modulation pathway, we analyzed the KEGG pathway involving the eight candidate genes using KOBAS (KEGG orthology-based annotation system, Fig. 3C). The relationship between TLR4 and IBS was meanwhile corroborated by previous literature (10). The binding site of miR-16 and TLR4 was unveiled from the microRNA.org (Fig. 3D) and confirmed by dual-luciferase reporter assay (Fig. 3E). TLR4-WT and miR-16 mimic cotransfection reduced the luciferase activity, while no significant difference was observed in luciferase activity between TLR4 MUT and miR-16 mimic cotransfection group and the mimic NC group. Therefore, miR-16 specifically binds to TLR4 mRNA on the translational level. The expression of TLR4, NF-κB p65, phosphorylated (p)-NF-κB p65 in the colorectal tissue of the IBS-D model was determined by Western blot (Fig. 3F), which were observed to be highly expressed in the IBS-D model. Further, the expression of miR-16, TLR4, NF-κB p65, and p-NF-κB p65 was measured by qRT-PCR and Western blot in response to mimic or inhibitor of miR-16 in the human normal colonic epithelial cell line NCM460 (Fig. 3, G and H). miR-16 was upregulated, while TLR4, NF-κB p65, and p-NF-κB p65 were downregulated in the presence of miR-16 mimic. In contrast, miR-16 inhibitor led to the opposite results. Thus, we confirmed that miR-16 inhibited the TLR4/NF-κB signaling pathway. Overexpression of miR-16, co-overexpression of miR-16 and TLR4, and overexpression of miR-16 with TLR4 inhibitor TAK-242 simultaneously were applied on the LPS-treated NCM460 cells. The results of qRT-PCR and Western blot (Fig. 3, I and J) suggested that miR-16 was elevated in the LPS-treated NCM460 cells transfected with miR-16 mimic while expression of TLR4, NF-κB p65, and p-NF-κB p65 was attenuated. Overexpression of TLR4 elevated the levels of TLR4, NF-κB p65, and p-NF-κB p65 in LPS-treated NCM460 cells transfected with miR-16 mimic. The addition of TAK-242 and miR-16 mimic presented higher levels of TLR4, NF-κB p65, and p-NF-κB p65 in LPS-treated NCM460 cells, as compared with miR-16 mimic alone. CCK-8 assay and flow cytometry (Fig. 3, K and L) indicated that transfection of miR-16 mimic augmented the cell viability and impeded the apoptosis, while further overexpression of TLR4 reversed the results. Relative to miR-16 mimic alone, simultaneous treatment of TAK-242 and miR-16 mimic promoted the NCM460 cell viability and repressed the apoptosis. Inflammatory cytokines were monitored by ELISA, qRT-PCR, and Western blot (Fig. 3, M and N). miR-16 mimic repressed the IL-1β and IL-6 and raised the level of ZO-1 and occludin, which could then be reversed by co-overexpression of TLR4. TAK-242 treatment in the LPS-treated NCM460 cells overexpressing miR-16 further repressed the IL-1β and IL-6 expression and elevated the level of ZO-1 and occludin. Subsequent TER detection indicated that the LPS-induced damage to tight junction integrity was relieved in response to miR-16 mimic, and the effect of miR-16 mimic alone was negated by additional TLR4 overexpression but potentiated when miR-16 mimic was combined with TAK-242 treatment (Fig. 3O). Together, miR-16 inhibited the TLR4/NF-κB signaling pathway to augment enterocyte viability, to inhibit apoptosis and cytokine production, and to maintain tight junction integrity.
We examined the XIST expression in the colorectal tissue of the IBS-D model using qRT-PCR (Fig. 4A), which revealed that XIST was highly expressed in the IBS-D. Treatment with NF-κB inhibitor pyrrolidine dithiocarbamate (PDTC) for 1 h decreased the XIST level, versus the dimethyl sulfoxide control (Fig. 4B). To explore the role of NF-κB and XIST in vitro, we treated the LPS-treated NCM460 cells with PDTC with or without XIST overexpression simultaneously. Western blot data (Fig. 4C) indicated that PDTC treatment (1 h) diminished the levels of NF-κB p65 and p-NF-κB p65 in LPS-treated NCM460 cells. The qRT-PCR analysis (Fig. 4D) showed XIST was downregulated in response to PDTC in LPS-treated NCM460 cells. CCK-8 assay and flow cytometry were then performed to detect the cell viability and apoptosis in LPS-treated NCM460 and CMEC cells in response to different treatments. According to the results, PDTC treatment (1 h) alone led to enhanced viability and attenuated apoptosis in the LPS-treated cells, while additional XIST overexpression reversed the effects of PDTC (Fig. 4, E–H). Then, cytokines and tight junction proteins were determined by ELISA, qRT-PCR, and Western blot (Fig. 4, I–K). It was found that PDTC augmented the cell viability, repressed the apoptosis, downregulated the IL-1β and IL-6, and elevated the expression of ZO-1 and occludin in LPS-treated NCM460 cells; whereas, additional XIST overexpression abolished the aforementioned effects of PDTC. Further detection of TER revealed that PDTC treatment (1 h) enhanced the tight junction integrity that had been damaged by LPS, while XIST overexpression abrogated the tight junction integrity-promoting effect of PDTC treatment (1 h) alone (Fig. 4L). In conclusion, inhibition of NF-κB suppresses XIST, thereby promoting enterocyte viability, inhibits apoptosis and cytokine production, and maintains tight junction integrity.
To investigate the effects of miR-16/TLR4/XIST on the NCM460 cells, we overexpressed miR-16, or together with XIST, in the LPS-treated NCM460 cells. The qRT-PCR (Fig. 5A) and Western blot (Fig. 5B) showed that miR-16 mimic transfection diminished expression of TLR4, NF-κB p65, p-NF-κB p65, and XIST in LPS-treated NCM460 cells. Further overexpressing XIST resulted in no remarkable changes in the level of miR-16, TLR4, and NF-κB p65. CCK-8 assay and flow cytometry were conducted to unveil the cell viability and apoptosis (Fig. 5, C and D), while cytokines and tight junction proteins were examined by ELISA, qRT-PCR, and Western blot (Fig. 5, E and F). miR-16 mimic alone resulted in increased cell viability, decelerated the apoptosis, suppressed the IL-1β and IL-6, and enhanced the expression of ZO-1 and occludin, and simultaneous overexpression of XIST reversed the effects of miR-16 overexpression alone. Consistently, the tight junction integrity-promoting effect of miR-16 mimic alone, as reflected by increased TER level, was reversed when miR-16 mimic was combined with XIST restoration (Fig. 5G). The aforementioned data demonstrated that miR-16 inhibited the TLR4/NF-κB pathway to repress XIST, enhanced enterocyte viability, repressed apoptosis and cytokine production, and maintained tight junction integrity.
To verify that miR-16 relieved IBS through the XIST axis, we overexpressed miR-16, or together with XIST, in the colorectal tissue of the IBS-D mice. The results of qRT-PCR and Western blot (Fig. 6, A and B) showed that overexpressing miR-16 suppressed the TLR4, NF-κB p65, and XIST in IBS-D mice. Further overexpression of XIST resulted in no notable difference in miR-16, TLR4, NF-κB p65, and p-NF-κB p65 levels but upregulated XIST expression in the presence of miR-16 overexpression. Next, we treated IBS-D mice with NF-κB p65 inhibitor and sh-XIST, respectively. According to the RT-qPCR and Western blot results, expression of XIST and NF-κB p65 was diminished in the NF-κB p65 inhibitor presence, and XIST was only reduced in sh-XIST presence (Fig. S1A). The results of ELISA showed that NF-κB p65 inhibitor inhibited the serum levels of IL-1β and IL-6 in IBS-D mice (Fig. S1B). In addition, H&E staining found that compared with the untreated IBS-D mice, the inflammatory cell infiltration was significantly reduced in the rectal mucosal tissues of NF-κB p65 inhibitor-treated IBS-D mice (Fig. S1C). Defecation intervals, stool types, and AWR scores of the mice were evaluated and summarized in Figure 6, C–E. IBS-D mice transduced with miR-16 agomir showed extended defecation intervals, decreased water content in stool, and raised AWR scores. Overexpression of XIST reversed symptoms aforementioned. Results of colorectal stepwise distention (Fig. 6F), EMG activity (Fig. 6G), and expressions of IL-1β and IL-6 (Fig. 6H) indicated increased injury threshold, eliminated EMG, and suppressed IL-1β and IL-6 in the IBS-D mice transduced with miR-16 agomir. Overexpressing the XIST level aggravated the symptoms of mice. H&E staining was conducted to evaluate the histological change of the intestinal mucosa (Fig. 6I). No inflammatory and other pathological damage in the rectum was observed. The results of TUNEL staining then showed that the number of apoptotic cells was reduced in the presence of miR-16 agomir alone, the effects of which could then be abrogated overexpression (Fig. 6J). All in all, we settled down with the conclusion that miR-16 inhibited the TLR4/NF-κB/XIST axis to relieve IBS-D.
IBS is a relapsing inflammatory disorder of the gastrointestinal tract that can lead to Crohn’s disease and ulcerative colitis (16). Evidence exists reporting that underexpression of miR-16 occurs in IBS-D, and miR-16 was involved in barrier function dysregulation through the modulation of Cldn2 and cingulin expression in IBS (6). Thus, efforts of our work are focusing on deciphering the molecular mechanisms underlying effect of miR-16 in the development of IBS-D. Our findings indicated that miR-16 inhibited the TLR4/NF-κB pathway to suppress XIST, augment intestinal epithelial cell viability, inhibit the apoptosis and inflammatory cytokines, and maintain tight junction integrity in IBS-D (Fig. 7). Understanding the regulation of miR-16 holds potential to efficient prevention and therapeutic strategies for IBS-D. Emerging evidence has pointed out the involvement of miRNAs in modulating specific biological processes in IBS-D, which may be translated clinically to restore intestinal functions and alleviate gastrointestinal symptoms (17, 18). This present study revealed that miR-16 was underexpressed in the IBS-D patients and mouse model. Corroborating finding has been identified in a previous study by Wohlfarth et al. (19), which demonstrated that miR-16 was poorly expressed in the jejunum of IBS-D patients. Further exploration unveiled that miR-16 overexpression enhanced enterocyte viability, restricted apoptosis and cytokine production, and maintained tight junction integrity in IBS-D. It is noted that deregulation of miR-16 expression is associated with intestinal disorders through mediating its target mRNAs (6). For a specific miRNA, the targets were various and have to be validated following in silico prediction. Herein, we predicted and verified TLR4 as a miR-16 target. A similar study has indicated the value of miR-16 in preventing acute lung injury through targeting TLR4 (20). Kocak et al. (21) have also found aberrant upregulation of TLR4 levels in IBS-D patients, which was observed to link with immune disorder along with oxidative stress. We moved on to explain the downstream mechanisms and found that the effect of miR-16 was realized by inhibiting the TLR4/NF-κB signaling pathway. A wide array of miRNAs has been proposed to modulate intestinal inflammatory reactions through the NF-κB signaling pathway. In IBS-related ulcerative colitis, miR-126 (22), miR-150 (23), and miR-155 (24) positively regulate and induce inflammation via the NF-κB signaling cascade. Consistent with our findings, it is reported that miR-16 targets and inhibits TLR4 in LPS-induced inflammatory pathway but the alteration is reversed by a lncRNA (25). He et al. (26) corroborated that expression levels of TLR4 and NF-κB were upregulated in IBS-D rats together with increased IL-8, TNFα, and myeloid differentiation factor 88 (MyD88), but the downstream effector of NF-κB was not identified. In parallel, it is documented that TLR4 activates NF-κB to upregulate cystathionine β synthetase and increase visceral hypersensitivity in an animal model of IBS (11). In relation to our findings, a number of previous studies have indicated the inhibitory effect of miR-16 on inflammation through TLR4/NF-κB signaling (20, 21, 27). For instance, miR-16 has been recognized as mediator of inflammatory responses through downregulating the transcription level of TLR4 and interleukin-1 receptor-associated kinase 1 (IRAK-1) (28). Further, miR-16 can downregulate the expression of NF-κB, NLRP3, and other inflammatory factors by targeting TLR4, thereby attenuating the inflammation in a LPS-induced acute lung injury model (20), and sepsis mice treated with miR-146a presented with decreased NF-κB activation as well as splenocyte apoptosis (29). Although the present study intersects with previous documentation in regard of miR-16 targeting TLR4/NF-κB signaling, it stands out for expanding the relatively well-recognized mechanism to IBS, indicating the promising potential of miR-16 in this new field. Further mechanistic investigations clarified that inhibition of NF-κB suppressed XIST to augment enterocyte viability, inhibit apoptosis and cytokine production, and maintain tight junction integrity. XIST is a 17 kb lncRNA located in the nucleus and maintains X chromosome inactivation (30). Prior evidence has suggested that silencing of XIST along the NF-κB pathway could halt LPS-induced inflammation in the lung (31). The mechanism underlying XIST expression in IBS-D has rarely been examined. We clarified that NF-κB positively related to the XIST level in IBS-D. Tight junction protein ZO-1 is deemed as an important scaffold protein, functioned as a barrier between the interior of the organism and the extracellular environment (32), and occludin was first identified in epithelial cells as an integral plasma membrane enzyme localized at the tight junction barrier. Increased level of occludin, together with ZO-1, is a sign of IBS-D mitigation. In IBS-D rats, miR-144 increases intestinal permeability and attenuates epithelial barrier function by directly targeting occludin and ZO-1 (33). It should also be noted that in this study, we used PDTC to inhibit NF-κB signaling pathway. Although PDTC is commonly applied as a NF-κB inhibitor (34, 35), it is not a NF-κB-specific inhibitor and has been reported to exert diversiform roles in a variety of cell bioactivities (36, 37, 38). Herein, whether there exist other PDTC-mediated molecular mechanisms contributing to enterocyte viability may need further investigations. Of note, our data demonstrated that overexpression of miR-16 elevated the level of occludin and ZO-1, a sign of improved tight junction in epithelial cells, the mechanism underlying which has not been well established but may be clarified from three aspects. First, another miRNA, miR-122a, has been suggested to modulate occludin degradation through tumor necrosis factor-α (TNF-α), thereby affecting the permeability of intestinal cells (39), and miR-16 has also been correlated with TNF-α (40). Therefore, miR-16 may regulate occludin and ZO-1 based on its interaction with TNF-α. On the other hand, miR-25-3p shuttled by exosomes in colorectal cancer can be transferred to vascular endothelial cells to mediate the expression of ZO-1 and occludin by targeting Kruppel-like factors (KLFs) (41), and miR-16 also serves as a regulator of KLF4 (42), so it may be speculated that miR-16 affects ZO-1 and occludin in a KLF4-dependent manner. Moreover, PKC isoforms have been implicated in the assembly of tight junctions for being able to regulate occludin and ZO-1 (43), and miR-16 has been related to PKCα, a PKC isoform (44). Hence, it is also possible that miR-16 regulates occludin and ZO through PKCα. Moreover, it has been previously documented that IL-1β precursor could be cleaved by LPS-induced protease and then be released outside the cell in the form of IL-1β (45). More recently, a report indicated that, in response to LPS induction, the release of functional IL-1β in microparticles was under a two-step regulation by GSDM-D and P2X7 (46), and it has further been revealed that IL-1β could be detected by ELISA assay in NCM460 cells exposed to LPS stimulation (47). The aforementioned evidence supports that IL-1β can be secreted out of the cells as a secreted protein under LPS induction. Ultimately, our in vitro and in vivo experiments have validated that miR-16 inhibited the TLR4/NF-κB pathway, thus repressing XIST to maintain intestinal epithelial tight junction integrity and to relieve IBS. Collectively, the evidence acquired in the present study showed that miR-16 targeted TLR4 to suppress NF-κB and the downstream XIST, resulting in accelerated enterocyte viability, tight junction integrity maintenance, and diminished inflammation and apoptosis. We believe that miR-16 may be a promising target for preventing or treating IBS-D as well as other intestinal diseases caused by manifold dysfunction. However, despite the involvement miR-16/TLR4/NF-κB axis in IBS-D, we have not elucidated the exact location of the miRNA. miR-16 may locate in the cytoplasm, extracellular vesicles, or extracellular matrix. Moreover, the current results do not imply that miR-16 is the only miRNA involved in the NF-κB pathway against IBS-D. Other miRNAs may target XIST, ZO-1, or other effectors in the pathway, which should be studied in future investigations. Future exploration using two signals to activate inflammation, including LPS and TNFa, may help further consolidating the model and understanding the mechanistic actions. Moreover, ongoing and future studies are required to expand validation of the effect of miR-16 on invasion of inflammatory cells in tissues in IBD animal models. Notwithstanding its limitations, this study does suggest the significance of miR-16 in IBS-D.
All the patients and healthy control participants signed informed consent documentation. The study was approved by the Ethics Committee of Changshu Hospital Affiliated to Nanjing University of Chinese Medicine and complied with the Declaration of Helsinki. Animal experiments were conducted with approval of the Animal Ethics Committee of Changshu Hospital Affiliated to Nanjing University of Chinese Medicine and in accordance with the Guide for the Care and Use of Laboratory animals published by the US National Institutes of Health.
The downstream genes of miR-16 were predicted by RNAInter. The genes were screened by the evaluation scores. Gene interactions were analyzed by STRING and visualized using Cytoscape 3.5.1. The core gene was chosen for further evaluation. IBS-related genes were searched in the GeneCards database using the keyword “irritable bowel syndrome” and plotted with the target core genes using jvenn. To further investigate the gene modulation pathway, we utilized KOBAS 3.0 for the KEGG pathway enrichment analysis. microRNA.org was adopted for analysis of the binding site for miR-16 and target genes.
Patients who underwent colonoscopy examination from Changshu Hospital Affiliated to Nanjing University of Chinese Medicine from March 2016 to March 2017 were enrolled in this study, including 37 IBS-D patients and 37 healthy controls. The demographics of IBS-D patients and healthy controls are shown in Table S1. IBS-D patients were included according to the ROME III standard: (1) recurrent abdominal pain or abdominal discomfort, for at least 3 days per month for the last 3 months, accompanied by the following: bowel pain or discomfort relief after defecation, changes in defecation frequency, changes in stools characterization; (2) abdominal pain or discomfort at least twice per week. (3) IBS-D scores according to the Bristol stool chart, defecating fluffy and mushy stools for more than 25% of the period while hard and lumpy stools for less than 25% of the period. Normal subjects who were negative for the colonoscopy and patients coexisted other diseases were excluded. Peripheral blood samples of the patients were collected.
A kit (Norgen Biotek Corp, NGB-55500) was used for this experiment. Briefly, 250 μl of serum samples were mixed with 750 μl of TRIzol LS reagent and left to stand for 5 min. The mixture was then added with 0.2 ml of chloroform, shaken vigorously for 15 s, and allowed to stand at room temperature (RT) for 2 min. Next, the mixture was centrifuged at 13,000 rpm for 5 min and the supernatant was harvested, 0.5 ml of which was pipetted into a new 1.5 ml EP tube and then mixed with 0.3 ml isopropyl alcohol, put into the adsorption column, centrifuged at 13,000 rpm for 15 s. Thereafter, the column was added with 500 μl washing buffer and centrifuged at 13,000 rpm for 15 s, which was repeated again. The adsorption column was put back into the centrifuge and centrifuged at 13,000 rpm for 2 min, with the residual ethanol thoroughly shaken off. The adsorption cartridge was placed into a 1.5 ml nuclease-free collection tube, where 20 to 50 μl nuclease-free H2O was supplemented and allowed to stand at RT for 2 min, followed by centrifugation at 13,000 rpm for 1 min. The eluate was the RNA product and cryopreserved. Finally, the RNA concentration was determined using a NanoDrop spectrophotometer.
Total RNA was extracted using TRIzol reagent and reversely transcribed into complementary DNA (cDNA) using TaqMan MicroRNA Assays Reverse Transcription primer (4427975, Applied Biosystems) and PrimeScript RT kit (Takara). Meanwhile, a PolyA tailing kit (B532451, Sangon Biotech) was utilized to generate cDNA from miRNA. qRT-PCR was performed with Fast SYBR Green PCR kit (Applied Biosystems) and ABI 7500 RT-PCR system (Applied Biosystems). U6 and GAPDH served as internal standards. The 2−ΔΔCT method was used to calculate the relative expression of genes. Primer sequences are listed in Table S2.
NIH mice at a specific pathogen-free level, with both genders equally divided, were purchased from the Experimental Animal Center of Changshu Hospital Affiliated to Nanjing University of Chinese Medicine. The mice were randomized into four experimental groups and one control group, with ten mice in each group. The IBS-D mouse model was established by acetic acid instillation (48, 49). After fed for 7 days, the control mice were instilled with distilled water, while the model mice were instilled with acetic acid (0.5 ml per day). The model was evaluated after instillation for 14 days. The stool type and gastrointestinal transit time of each group were recorded. Colonic sensitivity test and AWR scores were used for visceral sensitivity assessment. Histologic changes in colon mucosa were detected using H&E staining. Animal experiments were conducted following the protocols approved by the Animal Care and Use Committee of Changshu Hospital Affiliated to Nanjing University of Chinese Medicine and following the National Institutes of Health guidelines. The mouse model was evaluated (50) as follows: (1) defecation time: each mouse was given 0.5 ml phenol red by intragastric administration. The time from gavage to red feces defecation was recorded. (2) Fecal water content: each mouse was given 0.5 ml phenol red by intragastric administration. Feces were collected for 2 h after gavage, and the wet weight was measured. The corresponding dry weight was measured after air dried. Fecal water content was calculated, that is, fecal water content = (wet weight − dry weight)/wet weight of feces × 100%. (3) Visceral sensitivity: the colorectal distension was performed by placing a balloon in the rectum. The airbag was placed 2 cm from the anus. After adaptation, the air was quickly injected into the airbag to a specified pressure (20, 40, or 60 mmHg) and maintained 20 s. Experiments were repeated five times at each pressure with an interval of 4 min. Evaluators, without the knowledge of grouping and air pressure, independently and synchronously observed the mice AWR and recorded the scores. Averaged AWR was calculated to indicate visceral sensitivity. AWR scales are as follows: AWR0: no remarkable behavior response; AWR1: occasionally head movement with an immobilized body; AWR2: mild abdominal muscle contraction without uplift; AWR3: abdominal muscle contraction with uplift; AWR4: pelvis uplift with an arched spine. The IBS-D mice were grouped into (1) IBS-D + agomir NC + oe-NC group (injected with adenovirus vector containing agomir negative control and adenovirus vector containing overexpression negative control), (2) IBS-D + miR-16 agomir + oe-NC group (injected with adenovirus vector containing miR-16 agomir and adenovirus vector containing overexpression negative control), and (3) IBS-D + miR-16 agomir + oe-XIST group (injected with adenovirus vector containing miR-16 agomir and adenovirus vector containing XIST overexpression). Adenovirus vectors with the titer of 2 × 108 PFU/ml were prepared and inoculated under the left axilla of the mice using a 1 ml syringe. Afterward, the mice were housed in a specific pathogen-free laboratory for the subsequent experiments.
The balloon was fixed on the pipe connected to an automatic expansion device (G&J Electronic) for colon dilatation. After lubrication, the balloon was placed into the distal colon of the mice with the top of the balloon 0.5 to 1 cm from the anus. The mice were confined to plastic containers and allowed to adaption for 15 to 20 min before testing. Stepwise distention in the colon was given from 0 to 60 mmHg with a step of 5 mmHg until the first contraction of the testicles, tail, or abdominal muscles, which was defined as the injurious visceral hypersensitivity pain threshold. Colon dilatation was repeated within 5 to 10 min. Stimulus intervals and the averaged pressure for each mouse were recorded. Lateral oblique EMG was examined to determine the visceral sensitivity of the IBS-D mice. The mouse was anesthetized with an i.p. injection of 50 mg/kg pentobarbital sodium. Two electrodes were implanted in the lateral oblique and externalized at the back of the head. The colorectal was dilatated to 20, 40, or 60 mmHg for 20 s and repeated after a 2 min interval. EMG was recorded on an electromyogram amplifier module EMG 100C (Biopac Systems).
The cell supernatant or tissue homogenizer was collected. ELISA kits were applied for quantification of IL-1β and IL-6 (BMS224-2 and BMS213-2, Thermo Fisher). Absorbance at 450 nm was recorded using a microplate reader (Wallac 1420 Multilabel, PerkinElmer).
Mouse rectal tissue was embedded with paraffin wax, sliced, treated with xylene I for 10 min, and stirred in xylene II for 10 min. The section was then rinsed in absolute ethanol/xylene (v/v 1/1) for 1 min, dewaxed and hydrated, stained with hematoxylin for 10 min, and differentiated in 0.25% hydrochloric acid alcohol for 3 s. Afterward, the slides were washed in ethanol for 1 min, stained with 0.5% eosin solution for 1 min, washed with 3% ethanol for 30 s before gradient alcohol dehydration, and being rendered transparent by absolute ethanol/xylene (1:1) for 1 min. The sections were rinsed in the xylene I and II for 5 min each, sealed with neutral gum in the ventilation cabinet, and observed under an optical microscope (Olympus).
Prepared slides of the colorectal tissue were treated with 0.1% Triton X-100 in PBS, subjected to 2 min ice bath, permeabilized, and then treated with 50 μl TUNEL detection solution, followed by 60 min incubation (37 °C) in the dark. After mounting with antifluorescence quenching solution, the slides were observed with a fluorescence microscope (FV1000, Olympus), with an excitation wavelength ranging from 450 to 500 nm and an emission wavelength from 515 to 565 nm. Cells presenting with red fluorescence were recognized as apoptotic cells.
Human normal colonic epithelial cell line NCM460 was purchased from Biobw (bio-108818) and cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) and 1% penicillin and streptomycin (Sigma–Aldrich) in the incubator (37 °C, 5% CO2). A human colonic mucosal epithelial cell line CMEC was purchased from Biobw (bio-73503) and cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS. Human embryonic kidney cells HEK-293T (CRL-11268, ATCC) were cultured in DMEM under the same condition. Cells in the logarithmic growth period were digested with trypsin and seeded in a 6-well plate with 1 × 105 cells per well. Cells were cultured in the complete medium for 24 h until the confluence reached about 75%. Transfection was performed using Lipofectamine 2000. After 48 h of transfection, the efficiency was examined by qRT-PCR. Cells were grouped into a mimic NC group, a miR-16 mimic group, an oe-NC group, an oe-TLR42 group, and an oe-XIST group. Plasmids for silencing were synthesized from GenePharma (pGPU6/Neo, C02003). Plasmids for overexpression (pCDNA3.1- FLAG-LPA2 in overexpression groups) were purchased from Miaolingbio (P1224) and those for miR-16 inhibitor and its corresponding NCs were from GenePharma. We established the LPS-induced IBS-D model in the normal colonic epithelial cell line NCM460 (15). LPS concentration was 2 μg/ml, which was used as the control for LPS induction. The final concentration of dimethyl sulfoxide (as control) and TLR4 inhibitor TAK-242 (Merck Millipore) was 1 μM and that of NF-κB inhibitor PDTC was 10 μM. The treatment time was 1 h.
Cells were digested and suspended after 48 h of transfection. The cell density was adjusted to 1 × 105 cells/ml, and 100 μl of cell suspension was seeded into 96-well plates. The cells were cultured overnight. The cell viability was tested at 0, 24, 48, and 72 h after inoculation using CCK-8 kit (Beyotime). For the measurement, 10 μl CCK-8 reagent was added to each well and the cells were incubated for 4 h before the absorbance at 450 nm was recorded. The cell growth curve was meanwhile plotted.
Cells were collected by centrifugation at 2000g for 5 min. The medium was discarded. The cells were washed with cold PBS twice, suspended in the binding buffer (1×, 400 μl) with AnnexinV-FITC (5 μl), and incubated at 4 °C for 15 min in dark. The mixture was further incubated at 4 °C for 5 min in the dark after propidium iodide (10 μl) was added. The specimens were assessed in flow cytometry (FACSCalibur, BD Bioscience) within 1 h. All experiments were conducted independently in triplicate.
The cells were digested by trypsin, lysed with enhanced radioimmunoprecipitation assay lysis buffer containing trypsin (Boster Biotech), and quantified using bicinchoninic acid kit (Boster Biotech). The proteins were separated by SDS-PAGE and transferred onto polyvinylidene fluoride membrane. The membrane was blocked in 5% bovine serum albumin for 2 h and then incubated at 4 °C overnight with diluted primary antibodies: rabbit anti-ZO-1 (ab216880, 1:1000, Abcam), anti-occludin (ab167161, 1:5000, Abcam), anti-TLR4 (ab13556, 1:500, Abcam), anti-NF-κB p65 (ab16502, 1:2000, Abcam), anti-NF-κB p65 (phospho S536, ab76302, 1:1000, Abcam), and anti-GAPDH (ab181602, 1:5000, Abcam). After incubation, horseradish peroxidase (or AF488) labeled anti-rabbit IgG (ab6721, or ab150117, 1:2000, Abcam) was added for another 1 h incubation at RT. The immunoblots were visualized with enhanced chemiluminescence reagents (EMD Millipore). The images were captured and analyzed by ImageJ 1.48u (Bio-Rad). GAPDH was served as the internal reference. All experiments were conducted independently in triplicate.
The binding sites of miR-16 and TLR4 were predicted by starBase. HEK293T cells were cultured in DMEM containing 10% FBS at 37 °C, 5% CO2. The TLR4 3′-UTR cDNA containing miR-16 binding site (TLR4-Wt) was inserted into the pmirGLO vector. The TLR4 3′-UTR containing mutant binding site (TLR4-Mut) was synthesis by point mutation and inserted into the pmirGLO vector. The vectors were verified by sequencing, performed by Ribobio. The two pmirGLO reconstruction vectors were transfected into HEK293T cells using liposomes together with miR-16 mimic or NC mimic. The cells were incubated for 48 h before collection and lysis. The suspension (100 μl) was mixed with Renilla luciferase detection reagent (100 μl) to check the activity of Renilla luciferase. Similarly, the Firefly luciferase activity was tested on a SpectraMax M5 (Molecular Device) for 10 s with an interval of 2 s.
The TER of monolayer cells were measured with the STX2 double-rod EVOM voltage resistance meter. When the NCM460 cells formed a monolayer epithelial cell barrier and the measured TER value reached a stable level, the TER of each monolayer was determined with three replicate wells for each experimental group. The measured TER level was expressed as a percentage of the initial measured level.
Statistical data were processed by SPSS 21.0 (IBM Corp). Measurement data were presented as mean ± SD. Data comparison between two groups was performed by unpaired t test. Comparison of data among multiple groups was performed by one-way ANOVA with Tukey’s post hoc test. Statistical analysis concerning time-based measurements within each group was realized using repeated measures ANOVA with Bonferroni’s post hoc test. A value of p < 0.05 indicated a significant difference.
The original contributions presented in the study are included in the article/supplementary material; further inquiries can be directed to the corresponding authors.
This article contains supporting information.
The authors declare that they have no conflicts of interest with the contents of this article. |
|
PMC9647539 | Siyuan Zhao,Vincenzo Carnevale,Matthew Gabrielle,Eleonora Gianti,Tibor Rohacs | Computational and functional studies of the PI(4,5)P2 binding site of the TRPM3 ion channel reveal interactions with other regulators | 28-09-2022 | TRPM3,TRP channel,ion channel,phosphoinositide,computational modeling,fcTRPM8, flycatcher apo TRPM8,hM2, human M2 muscarinic,hTRPM3, human TRPM3,MM/GBSA, molecular mechanics/generalized Born surface area,mTRPM3α2, mouse TRPM3α2,PDB, Protein Data Bank,PI(4,5)P2, phosphatidylinositol 4,5-bisphosphate,PregS, pregnenolone sulfate,TEVC, two-electrode voltage clamp,TRPM3, transient receptor potential melastatin 3,TRPV1, transient receptor potential vanilloid 1,VMD, Visual Molecular Dynamics | Transient receptor potential melastatin 3 (TRPM3) is a heat-activated ion channel expressed in peripheral sensory neurons and the central nervous system. TRPM3 activity depends on the membrane phospholipid phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2), but the molecular mechanism of activation by PI(4,5)P2 is not known. As no experimental structure of TRPM3 is available, we built a homology model of the channel in complex with PI(4,5)P2via molecular modeling. We identified putative contact residues for PI(4,5)P2 in the pre-S1 segment, the S4–S5 linker, and the proximal C-terminal TRP domain. Mutating these residues increased sensitivity to inhibition of TRPM3 by decreasing PI(4,5)P2 levels. Changes in ligand-binding affinities via molecular mechanics/generalized Born surface area (MM/GBSA) showed reduced PI(4,5)P2 affinity for the mutants. Mutating PI(4,5)P2-interacting residues also reduced sensitivity for activation by the endogenous ligand pregnenolone sulfate, pointing to an allosteric interaction between PI(4,5)P2 and pregnenolone sulfate. Similarly, mutating residues in the PI(4,5)P2 binding site in TRPM8 resulted in increased sensitivity to PI(4,5)P2 depletion and reduced sensitivity to menthol. Mutations of most PI(4,5)P2-interacting residues in TRPM3 also increased sensitivity to inhibition by Gβγ, indicating allosteric interaction between Gβγ and PI(4,5)P2 regulation. Disease-associated gain-of-function TRPM3 mutations on the other hand resulted in no change of PI(4,5)P2 sensitivity, indicating that mutations did not increase channel activity via increasing PI(4,5)P2 interactions. Our data provide insight into the mechanism of regulation of TRPM3 by PI(4,5)P2, its relationship to endogenous activators and inhibitors, as well as identify similarities and differences between PI(4,5)P2 regulation of TRPM3 and TRPM8. | Computational and functional studies of the PI(4,5)P2 binding site of the TRPM3 ion channel reveal interactions with other regulators
Transient receptor potential melastatin 3 (TRPM3) is a heat-activated ion channel expressed in peripheral sensory neurons and the central nervous system. TRPM3 activity depends on the membrane phospholipid phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2), but the molecular mechanism of activation by PI(4,5)P2 is not known. As no experimental structure of TRPM3 is available, we built a homology model of the channel in complex with PI(4,5)P2via molecular modeling. We identified putative contact residues for PI(4,5)P2 in the pre-S1 segment, the S4–S5 linker, and the proximal C-terminal TRP domain. Mutating these residues increased sensitivity to inhibition of TRPM3 by decreasing PI(4,5)P2 levels. Changes in ligand-binding affinities via molecular mechanics/generalized Born surface area (MM/GBSA) showed reduced PI(4,5)P2 affinity for the mutants. Mutating PI(4,5)P2-interacting residues also reduced sensitivity for activation by the endogenous ligand pregnenolone sulfate, pointing to an allosteric interaction between PI(4,5)P2 and pregnenolone sulfate. Similarly, mutating residues in the PI(4,5)P2 binding site in TRPM8 resulted in increased sensitivity to PI(4,5)P2 depletion and reduced sensitivity to menthol. Mutations of most PI(4,5)P2-interacting residues in TRPM3 also increased sensitivity to inhibition by Gβγ, indicating allosteric interaction between Gβγ and PI(4,5)P2 regulation. Disease-associated gain-of-function TRPM3 mutations on the other hand resulted in no change of PI(4,5)P2 sensitivity, indicating that mutations did not increase channel activity via increasing PI(4,5)P2 interactions. Our data provide insight into the mechanism of regulation of TRPM3 by PI(4,5)P2, its relationship to endogenous activators and inhibitors, as well as identify similarities and differences between PI(4,5)P2 regulation of TRPM3 and TRPM8.
Transient receptor potential melastatin 3 (TRPM3) is a heat-activated, outwardly rectifying, and Ca2+-permeable nonselective cation channel expressed in a variety of tissues, including peripheral sensory neurons of the dorsal root ganglia, and the central nervous system (1). Its chemical activators include the endogenous neurosteroid pregnenolone sulfate (PregS) (2) and the synthetic compound CIM0216 (3). TRPM3 activity can be inhibited by a number of compounds, including natural flavonones, such as isosakuranetin (4), the nonsteroid anti-inflammatory drug diclofenac, and the antiepileptic medication primidone (5). TRPM3 is a very well-established peripheral noxious heat sensor. Genetic deletion of this channel in mice results in impaired noxious heat sensation (6, 7, 8) and impaired inflammatory thermal hyperalgesia (6, 8). TRPM3 inhibitors also reduce thermal hyperalgesia and basal heat sensitivity (4, 5, 8). Activation of Gi-coupled receptors inhibits TRPM3 activity. This effect was demonstrated both by native receptors in dorsal root ganglia neurons, including μ-opioid and GABAB receptors (9, 10, 11), as well as by heterologously expressing Gi-coupled receptors such as M2 muscarinic receptor and μ-opioid receptors (9, 10). Inhibition by Gi-coupled receptors is mediated by direct binding of Gβγ to the channel protein (9, 10, 11) through a short α-helical peptide encoded by an alternatively spliced exon in TRPM3, the costructure of which with Gβγ has been recently determined by X-ray crystallography (12). (TRPM3 has a large number of splice variants, and some of the alternatively spliced exons are in the N terminus (1), which makes residue numbering confusing (13, 14, 15)). The Gβγ binding peptide is present in TRPM1, the closest relative of TRPM3, which is also inhibited by Gβγ (16), but it is missing from the rest of the TRPM family. Activation of recombinant (9) or native (17) Gq-coupled receptors may also inhibit TRPM3, which is also mediated mainly by Gβγ binding (9). It was recently shown that mutations in TRPM3 are associated with developmental and epileptic encephalopathies manifesting as intellectual disability and seizures in children (13). The originally described two disease-associated mutations both showed a gain-of-function phenotype with increased basal activity and increased heat and agonist sensitivity (14, 15). This points to the importance of TRPM3 in the brain, but knowledge on the functional role of TRPM3 in the central nervous system is quite limited (18). Phosphoinositides, especially PI(4,5)P2, are common ion channel regulators (19, 20). Most TRP channels, including TRPM3 (21, 22), are positively regulated by phosphoinositides (23), but in some cases such as transient receptor potential vanilloid 1 (TRPV1) (24, 25), or transient receptor potential canonical channels (TRPC) (26, 27), this regulation is complex, and sometimes controversial, with both negative and positive effects having been proposed. With the exception of TRPM1, which is very difficult to study in expression systems, all members of the TRPM subfamily have been shown to be positively regulated by PI(4,5)P2, and no negative regulation has been proposed for any TRPM subfamily member (23). While cryo-EM structures are available for five of eight members, the only TRPM channel for which the PI(4,5)P2 binding site is revealed by structural studies is TRPM8 (28). Currently, it is not known which residues in the TRPM3 protein PI(4,5)P2 binds to, and there is no experimentally determined structure available for TRPM3. To fill this key gap in knowledge, we generated a homology model of TRPM3, based on the experimental structure of TRPM4 in the ligand-free (apo) state (29). We then docked PI(4,5)P2 to our model of TRPM3 and identified putative PI(4,5)P2-interacting residues in the pre-S1 segment, the S4–S5 linker, and the proximal C-terminal TRP domain. We validated our results by docking PI(4,5)P2 to an apo structure of TRPM8 (30), which showed remarkable similarity to the TRPM8–PI(4,5)P2 structures (28), experimentally determined recently. In silico mutations of the PI(4,5)P2 contact residues in TRPM3, followed by ligand-binding affinity changes via molecular MM/GBSA, showed reduced PI(4,5)P2 binding affinity to TRPM3. We experimentally validated the importance of these residues by demonstrating that their mutations increased sensitivity to inhibition by PI(4,5)P2 depletion in electrophysiology experiments. We also showed that mutating most of these residues increased sensitivity to Gβγ inhibition, and decreased sensitivity to agonist activation, indicating allosteric interaction between PI(4,5)P2 and endogenous activators and inhibitors. Furthermore, we demonstrated that gain-of-function disease–associated mutations did not change PI(4,5)P2 sensitivity, indicating that the mutations do not increase channel activity via promoting PI(4,5)P2 activation. Our data provide mechanistic insights into regulation of TRPM3 by its key endogenous cofactor PI(4,5)P2.
Our goal in this study was to identify the PI(4,5)P2 binding site of TRPM3. As there is currently no experimentally determined TRPM3 structure available, we generated a homology model of the human TRPM3 (hTRPM3) based on the cryo-EM structure of the mouse TRPM4 in the apo state (Protein Data Bank [PDB] ID: 6BCJ) (29) (Fig. 1, A and B). The template was selected as the closest homolog to TRPM3 in the TRPM family with an experimental structure available when the model was built (see the Experimental procedures section and Scheme S1 for details). Our model of TRPM3 aligns very well with the models of TRPM3 from different organisms generated recently by AlphaFold (31, 32) (Fig. S1), as well as with a model of TRPM3 obtained using the experimental structure of mouse TRPM7 (33) in EDTA (PDB ID: 5ZX5) as the template (Fig. S2), all of which became available after our original homology model was built, providing a posteriori validation of our TRPM3 model. Next, we identified putative residues interacting with PI(4,5)P2 in TRPM3 by using two complementary approaches. First, we scanned the surface of apo TRPM3 (model built on TRPM4) for putative binding sites using the program SiteMap (Schrödinger, LLC, 2018) (34, 35). Second, we relied on sequence and structural information available on TRPM8 to detect, by homology, which residues are likely to interact with PI(4,5)P2. Specifically, (1) we generated a sequence alignment of TRP-domain residues, K995, R998, and R1108, in the rat TRPM8, which are conserved among TRPM family members (Fig. 1C), and were previously suggested to play a key role in PI(4,5)P2 interactions (36) and (2) starting from the apo cryo-EM structure of the flycatcher apo TRPM8 (fcTRPM8) (PDB ID: 6BPQ) (30), we built a refined model of this channel bound to PI(4,5)P2 at a site that includes the conserved TRP-domain residues (Fig. 2, A and B). Comparing this complex with the apo-TRPM3 model showed that the most suitable site for lipid binding (i.e., the top-scoring binding spot combining SiteMap predictions and structural information) in TRPM3 corresponded to the PI(4,5)P2 site identified in TRPM8. We used this lipid-binding site to generate a model of TRPM3 in complex with a version of PI(4,5)P2 with truncated tails (similar to the synthetic diC8 PI(4,5)P2, which is fully functional in experiments) by molecular docking using the program Glide (37). We ranked the lipid binding modes by the standard precision scoring function. The best binding mode in TRPM3, defined as the best docking score (kilocalorie/mole) obtained at the binding site similar to that in TRPM8, is shown in Figure 1, A and B. The PI(4,5)P2 binding site in TRPM3 is formed by parts of the preS1 segment, the S4–S5 loop, and the proximal C-terminal TRP domain of the same subunit. The closest contact residues with PI(4,5)P2 are W761 in the preS1 segment, the N991 and K992 residues in segment connecting the voltage sensor–like domain (S1–S4) to the S4–S5 linker and the R1131 in the TRP domain (Fig. 1B). Figure 1B also shows the location of two additional residues, K1128 and R1141 in the TRP domain, which are not in close contact with PI(4,5)P2, but we experimentally characterized their mutations (see later). The numbering of these residues corresponds to the splice variant of hTRPM3 (hTRPM1325) (15, 38), which we used in the majority of our experiments. Comparing our model of TRPM8 in complex with PI(4,5)P2 with the subsequently determined two cryo-EM structures of TRPM8 with PI(4,5)P2 (28), icilin (PDB ID: 6NR3), and with the menthol analog WS12 (PDB ID: 6NR2), offered a posteriori validation of our modeling (Table S1). Fig. S3 compares the PI(4,5)P2 binding pockets of the TRPM8–PI(4,5)P2–icilin structure, the TRPM8–PI(4,5)P2–WS12 structure, and our computational model. Our model superposes very well with both structures in the transmembrane domains, but it shows a better structural alignment of the PI(4,5)P2 binding site with the TRPM8–PI(4,5)P2–WS12 (PDB ID: 6NR2) structure than the TRPM8–PI(4,5)P2–icilin–calcium structure (PDB ID: 6NR3). In fact, the minimum RMSD values, calculated over amino-acid ranges facing the lipid-binding sites between any two aligned structures, were 1.49 and 2.34 Å, respectively (Table 1). Interestingly, the structural difference between the two experimental structures (minimum RMSD of 1.99 Å) is larger than that observed between our model and the closest experimental complex (minimum RMSD of 1.49 Å). Yin et al. (28) listed five key residues in their cryo-EM costructures critical for PI(4,5)P2 interaction: R997 in the TRP domain, R850 in the S4–S5 loop, N692 and R688 in the pre-S1 segment (Fig. 1C), and K605 in the neighboring N-terminal cytoplasmic Melastatin Homology Region 4 (MHR4) domain. All these residues, with the exception of R850, are in contact with, or very close to PI(4,5)P2 in our TRPM8–PI(4,5)P2 model. R850 is in contact with the acyl chain of PI(4,5)P2 in our model, and only in contact with the PI(4,5)P2 headgroup in the 6NR3, but not in the 6NR2 structure, which is consistent with the better alignment of our model with the 6NR2 PI(4,5)P2-TRPM8 structure. Overall, our TRPM8–PI(4,5)P2 docking model validates our computational approach to identify the TRPM3 PI(4,5)P2 binding site and suggests that PI(4,5)P2 likely binds to a site that is similar in TRPM3 and TRPM8. Furthermore, superimposition of our model to the experimental structure of TRPM7 in EDTA (PDB ID: 5ZX5) (33) revealed that the docked PI(4,5)P2 in our model of TRPM3 fits well in a cavity of the experimental structure of TRPM7 that accommodates a detergent cholesteryl hemisuccinate molecule (Fig. S4). Whether this binding site is occupied by PI(4,5)P2 in TRPM7 in a cellular environment, remains to be determined, nevertheless the presence of this lipid-binding pocket in TRPM7 suggests that the location of the PI(4,5)P2 binding site may be conserved in multiple members of the TRPM subfamily. In our TRPM3-PI(4,5)P2 model, residues K992 and R1131 are equivalent to the experimentally determined PI(4,5)P2 contact sites R850 and R997 in TRPM8, located in the S4–S5 loop and the TRP domain (Fig. 1C). Specifically, N991 is adjacent to K992, and W761 in the pre-S1 segment of TRPM3 is shifted six residues from the R688 residue in TRPM8. The equivalent of W761 in TRPM8 (W682) is relatively close to PI(4,5)P2 in TRPM8, and so is the equivalent of R688 in TRPM3 (M767) highlighting the generally similar importance of the pre-S1 segment in PI(4,5)P2 binding in the two channels. The largest difference between the two binding sites is that the MHR4 region, which carries K605 in TRPM8, is not conserved in TRPM3, and the equivalent residue is far away from PI(4,5)P2 in our TRPM3 model. Overall, the two channels bind PI(4,5)P2 in a similar, yet not identical manner (Fig. S5). Next, we mutated the predicted PI(4,5)P2-interacting residues in TRPM3 and tested the effects of the mutations on sensitivity to inhibition by decreasing PI(4,5)P2 levels. We expressed the WT and mutant channels in Xenopus oocytes and performed two-electrode voltage clamp (TEVC) experiments. We stimulated channel activity with 50 μM PregS and measured current amplitudes, then incubated the oocytes with 35 μM wortmannin for 2 h, and measured PregS-induced currents in the same oocytes (Fig. 3, A and B). Wortmannin at this concentration inhibits phosphatidylinositol 4-kinases and has been used to inhibit the activity of PI(4,5)P2-dependent ion channels (39). We showed earlier that at 35 nM, a concentration that selectively inhibits phosphoinositide 3-kinases (PI3K), wortmannin did not inhibit TRPM3 (21), indicating that TRPM3 inhibition by 35 μM wortmannin is caused by inhibition of phosphatidylinositol 4-kinase, not PI3K. Mutating a PI(4,5)P2-interacting residue is expected to increase inhibition by high concentrations of wortmannin (39). We mutated the TRP domain positively charged residues to Q, as equivalent mutations in TRPM8 were shown to be functional, and affect PI(4,5)P2 interactions (36). The rest of the residues we mutated to A, but the W761A mutant was nonfunctional, thus we functionally characterized W761F instead. Mutations of all computationally predicted PI(4,5)P2-interacting residues (W761F, N991A, K992A, and R1131Q) showed significantly higher inhibition after wortmannin treatment than WT TRPM3 (Fig. 3, C–F), and their current amplitudes were also significantly lower than WT TRPM3 (Fig. 3, H–K). We also generated two additional mutations in the TRP domain in residues that are not in contact with PI(4,5)P2, K1128Q and R1141Q. Both mutants showed similar current amplitudes to WT TRPM3 (Fig. 3, K and L). The K1128Q mutant showed similar inhibition to WT (Fig. 3G), but the R1141Q mutant showed a small but significant increase in wortmannin inhibition compared with WT (Fig. 3F). This mutation is equivalent to R1008Q in the rat TRPM8, which reduced both PI(4,5)P2 and menthol sensitivity (36), but it was in contact with the menthol analog WS12, but not with PI(4,5)P2 in the cryo-EM structure of fcTRPM8 (R1007) (28). Therefore, it is possible that this mutation affected PI(4,5)P2 sensitivity indirectly. Next, we confirmed our data in whole-cell patch-clamp experiments using the rapamycin-inducible 4′ 5′ phosphoinositide phosphatase pseudojanin (40). Fig. S6, A–C shows that 100 nM rapamycin induced a significantly higher inhibition of the N993A mutant of the mouse TRPM3α2 (mTRPM3α2; equivalent of N991A in hTRPM3) than the WT mTRPM3α2 when the channels were stimulated with 25 μM PregS. Next, we stimulated the mTRPM3α2 with the combination of 25 μM PregS and 10 μM clotrimazole, which was shown to open an alternative pore, characterized by larger currents and less prominent outward rectification (41). Application of 100 nM rapamycin induced a significantly larger inhibition of currents induced by clotrimazole plus PregS in the N993A mutant compared with the WT mTRPM3α2 (Fig. S6, E–G). Current amplitudes for the N993A mutant were also lower than those in the WT TRPM3 (Fig. S6, D and H). Mutation of the PI(4,5)P2 contact site R998Q resulted in a right shift in the diC8 PI(4,5)P2 dose response in excised patches (36). TRPM3 currents in excised patches show a less steep concentration dependence, with no clear saturation at higher concentrations (21). This and the low current amplitudes in the mutants prevented us from reliably comparing PI(4,5)P2 dose responses in our mutants. It was reported for TRPV1 that mutating a putative PI(4,5)P2-interacting residue increased the relative efficiency of PI(4)P to stimulate channel activity compared with PI(4,5)P2 (42). Therefore, we tested the relative effects of PI(4)P and PI(4,5)P2 on the N991A mutant. Fig. S7, A–C shows that the relative effect of PI(4)P compared with PI(4,5)P2 did not change. Next, we used MM/GBSA calculations (Fig. 4) to predict the changes in the binding free energy (ΔΔG) of PI(4,5)P2 to the native (WT) TRPM3 versus the mutant channels that were characterized in Figure 3. In particular, we used the VSGB 2.0 model (43), in which the solvation free energy is approximated with an optimized model based on the surface generalized Born method (44) and the variable dielectric treatment of polarization (45) for protein residues. We note that we did not include an implicit membrane model (i.e., a low-dielectric slab) and, therefore, the results should be taken as a qualitative indication. As shown in Figure 4, the binding of PI(4,5)P2 is guided by a number of stabilizing interactions (Fig. 4A) established with key contact residues (Fig. 4, B–F and Table S2). Mutations of all these residues in our model resulted in a decreased PI(4,5)P2 binding affinity (for a native protein to bind better than the mutant, the calculated ΔΔG value is positive). Specifically, K992A had a more prominent effect than R1131Q, N991A, and W761F, respectively. This correlates well with K992A also having the most pronounced effect on inhibition by PI(4,5)P2 depletion (Fig. 3E). Regarding the binding modes, K992 engages in multiple interactions with PI(4,5)P2, including three hydrogen bonds and three salt bridges. Mutating K992 to alanine resulted in the loss of all these interactions with the exception of one hydrogen bond, the only interaction established by the amino acid backbone (Fig. 4C). Similar behavior was observed with the mutation R1131Q, the contact residue exerting the second largest effect on the binding affinity, which resulted in the loss of one hydrogen bond and two salt bridges (Fig. 4D), all established by the residue side chain. Next, mutating N991 to alanine and W761 to phenylalanine resulted in the loss of one hydrogen bond each (Fig. 4, E and F, respectively). To further corroborate our results, we performed additional sets of calculations of the binding affinity change upon mutation (Table S2 and Fig. S8) using, as the starting configurations, the docking poses of PI(4,5)P2 with even shorter tails than the ones included in the model, and with headgroups featuring different protonation states (hereinafter referred to as shortest-PI(4,5)P2). These calculations are clearly reproducible and agree with experimental observations. Although the trend is maintained overall (Figs. 4A and S8), W761F shows a reduction in the binding affinity change for shortest-PI(4,5)P2 (Fig. S8, light blue), because of headgroup protonation states that prevent interactions via hydrogen-bond formation. Hence, it appears from our calculations, that the PI(4,5)P2 protonation state featured in the proposed TRPM3 model (Fig. 4) is the one that favorably affects the binding of the phospholipid to the native protein. Interestingly, the protonation state of PI(4,5)P2 was suggested to critically impact the binding to related TRP channels (46). Furthermore, among the mutations leading to a decrease of binding affinity, W761 is located the furthest from PI(4,5)P2 (see structural model), and therefore, it is not unexpected that mutating this residue could affect to a lesser extent the binding of a smaller ligand (Table S2). Of the remaining two mutations (Fig. 4, A and B), R1128Q had only a very small effect on both ΔΔG and the related binding mode, which correlates well with it not being in close contact with PI(4,5)P2 in our model, and the lack of effect on wortmannin inhibition. The R1141Q mutant, which is also not a PI(4,5)P2 contact site, also had only a minimal effect on both ΔΔG and the related binding mode, indicating that the small, but significant, effect on wortmannin inhibition was likely because of indirect effects. Overall, all Δ affinity calculations supported our computational docking and agreed with the experimental functional characterization of the PI(4,5)P2-interacting residues. It is worth mentioning that, although our binding model likely captures a highly represented conformational state sampled by the TRPM3 channel when bound to PI(4,5)P2, it is expected that other states may exist featuring alternative networks of interactions yet compatible with the proposed phospholipids-binding site model. Next, we mutated two residues in the rTRPM8 that are equivalent to PI(4,5)P2-interacting residues in our TRPM3 model. The R851 residue in TRPM8 corresponds to the K992 residue in the S4–S5 linker in TRPM3 (Fig. 1C), and it was in direct contact with PI(4,5)P2 in the cryo-EM structure of the fcTRPM8 (R850) (28). The W682 residue is the equivalent of W761 in TRPM3 (Fig. 1C), and while is not in a direct contact with PI(4,5)P2 in the fcTRPM8 cryo-EM structure (R850), it is located relatively close. Since the W682A mutant was nonfunctional, we characterized the W682Q, which displayed small, yet measurable, menthol-induced currents. Figure 5, A–G shows that both the R851Q and the W682Q mutants showed significantly higher level of inhibition by wortmannin, with W682Q having a larger effect. Current amplitudes showed a similar pattern; both mutants were significantly lower than WT TRPM8, and the W682Q having a larger effect (Fig. 5H). The decrease in amplitudes was even more pronounced at negative voltages for inward currents (Fig. 5I), in agreement with earlier results with the R995Q PI(4,5)P2 mutant (36). This is likely caused by the allosteric interaction between PI(4,5)P2 and voltage in modulating TRPM8. Wortmannin treatment substantially accelerated deactivation after cessation of menthol stimulation (Fig. 5, A–D), which is in contrast to TRPM3, where the deactivation kinetics after washing out PregS was not affected by wortmannin (Fig. 3, A and B). Stimulation with menthol, or cold, was shown to increase the apparent affinity of TRPM8 for PI(4,5)P2 (36) indicating an allosteric interaction between menthol and PI(4,5)P2 activation. Next, we asked if this allosteric interaction also happens in the opposite direction and tested if mutations of the PI(4,5)P2-interacting residues in TRPM8 have an effect on agonist sensitivity. Figure 6, A–D shows that both the R851A and the W682Q mutant had right shifted menthol dose response. Similar to the effect on current amplitudes and wortmannin inhibition, the effect of the W682Q mutant (Fig. 6, C and D) was more pronounced than that of R851A (Fig. 6, B and D). We also tested if a similar allosteric effect also exists in TRPM3. Figure 7, A–D shows that both the N991A and the K992A mutant shifted the PregS dose response to the right. TRPM3 activity is inhibited by direct binding of Gβγ to the channel (9). To test if an allosteric interaction between Gβγ inhibition and PI(4,5)P2 activation is present, we expressed WT and mutant TRPM3 channels with or without Gβ1γ2 in Xenopus oocytes and measured PregS-induced currents. The N991A and K992A mutants were inhibited significantly more by Gβγ than WT TRPM3 (Fig. 8, A–E). The K1128Q mutant, which did not affect PI(4,5)P2 sensitivity, had no effect on Gβγ inhibition either (Fig. 8E). Interestingly, the R1131Q mutant was not inhibited, rather potentiated by coexpressing Gβγ (Fig. 8E). Consistently with the lack of inhibition by Gβγ, the R1131Q mutant was also not inhibited by stimulating Gi-coupled M2 muscarinic acetylcholine receptors (Fig. 8, F–J). These data indicate that while there is an allosteric interaction between PI(4,5)P2 and Gβγ, the R1131 residue in the TRP domain also plays some role in transmitting the inhibitory effect of Gβγ. Gain-of-function mutations in TRPM3 have recently been shown to cause intellectual disability and seizures (13, 14, 15). The two disease-associated mutations, V990M and P1090Q, were shown to increase basal channel activity, as well as increase in agonist sensitivity and increase in heat sensitivity, with V990M affecting agonist sensitivity more prominently, whereas P1090Q predominantly affecting heat sensitivity (15). Next, we tested if the increased basal activity and agonist sensitivity also translated into higher sensitivity to PI(4,5)P2. When WT (Fig. 9, A and B) and V990M (Fig. 9, C and D) and P1090Q mutant channels (Fig. 9, E and F) were treated with 35 μM wortmannin for 2 h, currents evoked by 50 μM PregS were inhibited to a similar extent (Fig. 9G). PregS-induced average current amplitudes were not significantly different in the mutant and WT channels (not shown), similar to our earlier data (15), presumably because the overactive channels tend to damage the cells expressing them and thus in the surviving oocytes are selected for lower expression levels of the mutants. The mutants were also inhibited to a similar extent to WT channels by wortmannin when currents were evoked by PregS corresponding to the respective EC50 (15) of the mutant and WT channels (Fig. 9H). These data indicate that the disease mutants do not increase channel activity by increasing their apparent affinity for PI(4,5)P2.
Our work aims to understand the molecular mechanism of PI(4,5)P2 regulation of TRPM3. We used computational docking, changes in the binding affinity estimated by computational mutagenesis, site-directed mutagenesis, and electrophysiology to identify PI(4,5)P2-interacting residues in the channel protein. Our data indicate that residues in three regions, the pre-S1 segment, the S4–S5 loop, and the TRP domain, play important roles in forming the PI(4,5)P2 binding site in TRPM3. Mutations of PI(4,5)P2-interacting residues decreased the binding affinity in silico (positive ΔΔG values in Figure 4) indicating that the native protein binds better than the mutant) and increased sensitivity to inhibition by decreasing PI(4,5)P2 levels in electrophysiology experiments (Fig. 3). Mutating PI(4,5)P2-interacting residues also decreased sensitivity to PregS activation and increased sensitivity to Gβγ inhibition indicating allosteric interaction between PI(4,5)P2 and agonists as well as a physiological inhibitor. On the other hand, disease-associated gain-of-function mutations did not change PI(4,5)P2 sensitivity, indicating that the mutations did not increase channel activity by enhancing PI(4,5)P2 activation. There are currently five channels in the TRPM family for which structural data are available (47): TRPM2 (48), TRPM4 (49), TRPM5 (50), TRPM7 (33), and TRPM8 (28). While all these channels have been shown to be positively regulated by PI(4,5)P2 (23), only TRPM8 has a costructure with this lipid (28). When we compare the PI(4,5)P2 binding site in TRPM8 revealed by the structural study with our computationally identified binding site in TRPM3, the two overlap, sharing some of the interacting residues, with some differences (Figs. 1C and S5). Overall, the preS1 segment, the S4–S5 loop, and the TRP domain are involved in both channels in forming the PI(4,5)P2 binding site. The R1131 residue in the TRP domain in TRPM3 is equivalent to the R997 PI(4,5)P2 contact residue in the fcTRPM8 structure (28), and to the R998 residue in the rat TRPM8, which was proposed as a PI(4,5)P2-interacting residue and experimentally shown to exhibit decreased PI(4,5)P2 sensitivity before structures became available (36). The K992 residue in the S4–S5 loop of TRPM3 is equivalent to the R850 PI(4,5)P2 contact residue in the fcTRPM8 structure and to the R851 residue in the rat TRPM8 that we characterized in this study (Figs. 5 and 6). The pre-S1 segment of the fcTRPM8 has two PI(4,5)P2 contact residues R688 and N692 (Fig. 1C). These residues are not conserved in TRPM3 (Fig. 1C); yet the equivalent residues in our TRPM3 model, that is, M767 and G672, are both located within 5 Å of the PI(4,5)P2 headgroup (Fig. S5D), with M767 engaging hydrophobic interactions that stabilize the overall complex. The W761 PI(4,5)P2 contact residue in the pre-S1 of TRPM3, is equivalent to the W682 residue in TRPM8, which is close to the R688 residue as well as to PI(4,5)P2, but it was not close enough to designate it as a PI(4,5)P2 contact site in TRPM8 (28). Interestingly, when we mutated this residue to a glutamine (W682Q) in the rat TRPM8, it behaved similar to the W761F mutation in TRPM3, that is, it increased sensitivity to PI(4,5)P2 depletion (Fig. 5). Whether this residue is in a closer contact with PI(4,5)P2 in a cellular environment in the rat TRPM8, or its mutation affected PI(4,5)P2 interactions indirectly, or both, it is difficult to tell. Finally, the K605 residue from an adjacent cytoplasmic MHR4 domain was also in contact with PI(4,5)P2 in TRPM8. This residue is not conserved in TRPM3 and was not close to PI(4,5)P2 in our homology model. It is well established that channel agonists can increase PI(4,5)P2 sensitivity (apparent affinity) for various PI(4,5)P2-sensitive ion channels. For example, the apparent affinity of the G protein–activated inwardly rectifying K+ channel GIRK4 (Kir3.4) for PI(4,5)P2 is increased by factors that stimulate channel activity, such as Gβγ and intracellular Na+ (51). The apparent affinity of TRPM8 for PI(4,5)P2 was shown to be increased by both cold and menthol (36), and the apparent affinity of TRPV1 for PI(4,5)P2 activation was increased by capsaicin (52). The opposite was also proposed, as a mutation in the putative PI(4,5)P2-interacting residue R1008 in TRPM8 not only decreased apparent affinity for PI(4,5)P2 but also induced a marked right shift in the menthol dose response (36). In the view of the structures of TRPM8 however, this residue is likely to be a menthol-interacting residue, as it was in close contact in the TRPM8 structure with the menthol analog WS12, but not with PI(4,5)P2 (28), therefore, it most likely primarily affected menthol sensitivity, and the effect on PI(4,5)P2 was a secondary allosteric effect. Our data indicate that in both TRPM8 and TRPM3, mutating PI(4,5)P2 contact residues also decrease agonist sensitivity. Similarly, mutating most PI(4,5)P2-interacting residues also made it easier for TRPM3 to be inhibited by Gβγ. This is likely to be an allosteric effect, as the Gβγ binding peptide in TRPM3 (12) is located far away from the PI(4,5)P2 binding site (Fig. S9). Interestingly, the R1131Q mutant did not display any Gβγ inhibition, pointing to the complex role of this residue in channel regulation. In contrast to the apparent allosteric interaction between PI(4,5)P2 and agonist or Gβγ, disease-associated gain-of-function mutations in TRPM3 that increased both heat and agonist sensitivity (15) did not decrease sensitivity for inhibition by PI(4,5)P2 depletion (Figure 9), indicating that the mechanism of increased channel activity is not the consequence of increased sensitivity to PI(4,5)P2. In an earlier work, well before structures became available, three residues in the TRP domain of TRPM8 were proposed to act as PI(4,5)P2-interacting residues (36). While mutations in all three of them decreased PI(4,5)P2 apparent affinity (36), only one of them was in direct contact with PI(4,5)P2 in the TRPM8-PI(4,5)P2 structures that were determined later (28). Also before TRP channel structures became available, a short “PH domain–like” segment with several positively charged residues was proposed to act as a PI(4,5)P2 interaction site in TRPM4 (53). Even though mutations in this segment behaved in a way compatible with reduced PI(4,5)P2 interactions, this segment was far away from the plasma membrane in the subsequently determined TRPM4 structures, which is incompatible with acting as a PI(4,5)P2-interacting domain (49). It was also proposed that similar, nonconserved, and short charged amino acid segments are responsible for the effects of PI(4,5)P2 on other TRP channels, including TRPM3 and TRPM8, but for channels other than TRPM4, no experimental testing was performed with mutants in those segments (54). The proposed segments are at different locations in different TRPM channels, suggesting that the activation mechanism by PI(4,5)P2 is not conserved between different TRPM channels. Our data showing that the PI(4,5)P2 binding site in TRPM3 is similar, yet not identical to that in TRPM8 suggests that the PI(4,5)P2 binding site in TRPM channels shows substantial level of conservation. In our earlier work, we used a homology model, based on the structure of TRPV1 combined with mutagenesis, to predict PI(4,5)P2-interacting residues in the epithelial Ca2+ channel TRPV6 (55). Our homology model–based PI(4,5)P2 site was very similar to the experimentally determined PI(4,5)P2 binding site in TRPV5, with the same key contact residues (55, 56). TRPV5 and TRPV6 are products of a relatively recent gene duplication, and they share 75% identity, and they are functionally far more similar to each other than to other members of the TRPV subfamily. This gives us confidence that our computationally determined PI(4,5)P2 binding site in TRPV6 likely reflects the actual PI(4,5)P2 binding site with a reasonable accuracy, even if there is no TRPV6 PI(4,5)P2 costructure available currently. Similarly, in the current work, we docked PI(4,5)P2 to the apo structure of TRPM8 (30), which showed a high level of overlap with the PI(4,5)P2 binding site of TRPM8 determined subsequently by cryo-EM studies (28). This makes us confident that our experimentally tested computational prediction of the PI(4,5)P2 binding site in TRPM3 reflects the functionally relevant PI(4,5)P2 binding site with reasonable accuracy. In conclusion, our data provide mechanistic insight into regulation of TRPM3 by its key physiological cofactor, PI(4,5)P2. We identify its binding site on the channel, characterize the interaction between PI(4,5)P2 and other physiological regulators of TRPM3, and compare its regulation by PI(4,5)P2 to that of TRPM8.
All procedures of preparing X. laevis oocytes were approved by the Institutional Animal Care and Use Committee at Rutgers New Jersey Medical School. Frogs were anesthetized in 0.25% ethyl 3-aminobenzoate methanesulfonate solution (pH 7.4) (MS222; Sigma–Aldrich), then bags of ovaries were surgically collected, and rotated with 0.1 to 0.3 mg/ml type 1A collagenase (Sigma–Aldrich) at 16 °C overnight in OR2 buffer containing 82.5 mM NaCl, 2 mM KCl, 1 mM MgCl2, and 5 mM Hepes (pH 7.4). Afterward, oocytes were washed with OR2 several times and then kept in OR2 solution supplemented with 1.8 mM CaCl2, 100 IU/ml penicillin, and 100 μg/ml streptomycin at 16 °C. To express exogenous proteins, RNA was microinjected into oocytes using a nanoliter-injector system (Warner Instruments). RNA was in vitro transcribed from the linearized pGEMSH vectors, which contained the complementary DNA clones for hTRPM3 (38) rat TRPM8, human M2 muscarinic (hM2) receptor, or Gβγ subunits by using the mMessage mMachine T7 Transcription Kit (Thermo Fisher Scientific). TRPM3 and TRPM8 mutants, which were used in this article, were generated by the QuikChange II XL Site-Directed Mutagenesis Kit from Agilent, and the mutated DNA constructs were confirmed by DNA sequencing. For coexpression of TRPM3 constructs and Gβ1γ2 subunits, 40 ng of TRPM3 was coinjected with 5 ng Gβ1 and 5 ng Gγ2. In the case of coexpressing TRPM3 and hM2 receptors, these two were injected at 1:1 ratio, 40 ng each. Oocytes were used for electrophysiological experiments after 48 to 72 h incubation at 16 °C after RNA injection.
Oocytes were placed in extracellular solution, which contained 97 mM NaCl, 2 mM KCl, 1 mM MgCl2, and 5 mM Hepes, pH 7.4. Currents were measured with a protocol consisting a voltage step from the 0 mV holding potential to −100 mV, followed by a ramp to 100 mV once every 0.5 s with a GeneClamp 500B amplifier and analyzed with the pClamp 9.0 software (Molecular Devices). Currents were recorded by thin wall glass pipettes that contained inner filament and were filled with 1% agarose in 3 M KCl. In all TEVC experiments, different concentrations of PregS were applied to activate TRPM3 channels, and various concentrations of menthol were used to trigger responses of TRPM8 channels. The hM2 receptor was activated by 5 μM acetylcholine. For wortmannin experiments specifically, PregS, or menthol-induced currents were measured, then the same oocyte was incubated with 35 μM wortmannin for 2 h, and currents were measured again using the same protocol. In the bar graphs in Figure 3, the individual panels show experiments that were performed on the same day.
Oocytes were placed in a recording chamber filled with bath solution, which contained 97 mM KCl, 5 mM EGTA, 10 mM Hepes, pH 7.4. Before starting measurements, the vitelline layer was carefully removed with forceps without damaging the oocyte. Then a giga-ohm seal was formed using a borosilicate glass pipette (World Precision Instruments) with resistance from 0.8 to 1 MΩ. The pipette was filled with a solution containing 97 mM NaCl, 2 mM KCl, 1 mM MgCl2, 5 mM Hepes, and 100 μM PregS at pH 7.4. Currents were measured by an Axopatch 200B amplifier and analyzed with the pClamp 9.0 software. Compounds were dissolved in the bath solution and delivered to the inner side of cell membrane by a custom-made gravity-driven perfusion system. Either 25 μM PI(4,5)P2, 25 μM PI(4)P, or 10 μM AASt PI(4,5)P2 was applied in these experiments to reactivate TRPM3. At the end of every recording, 30 μg/ml Poly-Lys (Poly-K) was applied.
Human embryonic kidney 293 cells were purchased from American Type Culture Collection (catalog number: CRL-1573). Human embryonic kidney 293 cells were cultured in minimum essential medium supplemented with 10% fetal bovine serum and 100 IU/ml penicillin plus 100 μg/ml streptomycin. Cells were incubated in 5% CO2 at 37 °C. Cells were tested to confirm that they were not infected by mycoplasma. Cells were used up to 25 passages and then discarded. Cells were transiently transfected with complementary DNA encoding different TRPM3 constructs (200–400 ng) using the Effectene reagent (Qiagen). mTRPM3α2 and its mutant were cloned into the bicistronic pCAGGS/IRES-GFP vector. The components of rapamycin-inducible pseudojanin phosphatases (40) were cotransfected with mTRPM3α2 at 1:1 ratio.
After 24 h of transfection, cells were plated on poly-d-lysine–coated 12 mm cover slips. Experiments were performed 48 to 72 h after transfection. Coverslips were placed in recording chamber filled with extracellular solution (137 mM NaCl, 5 mM KCl, 1 mM MgCl2, 10 mM Hepes, and 10 mM glucose, pH 7.4). Since mTRPM3 constructs were in the background of bicistronic pCAGGS/IRES-GFP vector and rapamycin-inducible phosphatases were labeled with red florescent protein, cells that showed both GFP and red florescent protein fluorescence were selected for the whole-cell patch-clamp experiments. Patch pipettes were prepared from borosilicate glass capillaries (Sutter Instruments) using a P-97 pipette puller (Sutter Instrument) with a resistance of 2 to 4 MΩ. Those recording pipettes were filled with intracellular solution containing 140 mM potassium gluconate, 5 mM EGTA, 1 mM MgCl2, 10 mM Hepes, and 2 mM Na-ATP, pH 7.4. After formation of gigaohm-resistance seals, the whole cell configuration was established, and currents were recorded by applying a ramp protocol once every 1 s. The holding potential was 0 mV; followed by a −100 mV step for 100 ms; plus a ramp protocol from −100 mV to +100 mV over the period of 500 ms. All recordings were made with an Axopatch 200B amplifier, filtered at 5 kHz, and digitized through a Digidata 1440A interface. Data were collected and analyzed with the pClamp10.6 (Clampex) acquisition software (Molecular Devices) and further analyzed and plotted with Prism 9 (GraphPad by Dotmatics). TRPM3 channels were activated by PregS, and 100 nM of rapamycin was applied to activate phosphatases.
Statistical analysis was performed with Origin 2021 and GraphPad Prism 9. Data were plotted as mean ± SEM and scatter plots or mean ± SD when scatter plots are not provided. Sample sizes were not predetermined by any statistical method; however, they were similar to what is generally used in the field. All recordings were performed in random order. Statistical significance was evaluated with t test, or ANOVA with Bonferroni’s post hoc test, or the Kolmogorov–Smirnov nonparametric test, using GraphPad Prism 9, as described in the figure legends. p Values are reported in figures or figure legends.
The cryo-EM structure of full-length apo TRPM8 from Ficedula albicollis (PDB ID: 6BPQ) (30), which contains several unresolved amino acid ranges (∼4.1 Å resolution) as well as protein residues with missing atoms, was used as the starting configuration to generate a refined structural model of the TRPM8 channel. The Prime Loop Prediction (57) program and the Protein Preparation Wizard (58) (both distributed by Schrödinger, LLC, 2018) were used to perform the following tasks: (1) loop refinement by serial loop sampling, at the ultraextended accuracy level. In particular, four unresolved amino acid ranges in the transmembrane region were sampled, including 714 to 722, 819 to 822, 889 to 895, and 976 to 990 (sequence numbering as in F. albicollis); (2) side-chain prediction of protein residues with missing atoms, performed with no backbone sampling; (3) pKa prediction of protein residues at pH 7, followed by analysis and optimization of hydrogen-bond networks; (3) structure refinement via restrained minimization of heavy atoms (hydrogens not restrained) using the OPLS (59) force field. The minimization convergence criterion was set to 0.30 Å RMSD for heavy atom displacement. The resulting apo TRPM8 structure was then searched for putative ligand-binding sites using SiteMap (35). Residues facing the topmost suitable site for ligand-binding spot were used to define the docking space for putative PI(4,5)P2 binding modes. The program Glide (60) (Schrödinger, LLC, 2018) was used to dock PI(4,5)P2 against TRPM8, using a rigid-receptor and flexible-ligand protocol. The ligand was prepared by using the default protocol of LigPrep (Schrödinger, LLC, 2018). Binding modes were ranked using the Glide standard precision scoring function. The best binding mode of PI(4,5)P2 against TRPM8 is shown in Figure 2. After our refined TRPM8–PI(4,5)P2 complex was generated and used for subsequent modeling of the TRPM3 channel as in a complex with PI(4,5)P2, seven additional experimental structures of TRPM8 became available (Table S1). Three of these structures report the TRPM8 channel in complex with PI(4,5)P2 as well as Ca2+ ions and/or small-molecule ligands (28).
No experimental structure of the TRPM3 channel is currently available. The cryo-EM structure of the TRPM4 channel (3.1 Å resolution) in the apo state with short coiled coil from Mus musculus (PDB ID: 6BCJ) (29) was selected as the template to build a homology model of the hTRPM3 structure using the Swiss-Model Server (61) (https://swissmodel.expasy.org/), based on the human sequence UniProtKB: Q9HCF6. The choice of the template is exemplified in Scheme S1. Essentially, the closest relative to TRPM3 in the TRPM family (cladogram) with an available structural template was selected, that is, TRPM4 (29). The cladogram was generated using Clustal Omega (https://www.ebi.ac.uk/), upon performing a multiple sequence alignment (default settings) (62). The Swiss-Model–generated protein structure of apo TRPM3 was then prepared for subsequent calculations using the Protein Preparation Wizard (58). Potential hot spots for PI(4,5)P2 binding to TRPM3 were defined by combining binding-site mapping results obtained using SiteMap (35) with sequence and structure alignments between the refined structural model of TRPM8 in complex with PI(4,5)P2 and the TRPM3 model (apo state) generated using Swiss-Model (61). Hence, the TRPM3 protein residues facing the most “druggable” binding spot were selected by homology and used to center the docking grid for subsequent docking of PI(4,5)P2. The best binding mode of a truncated version of PI(4,5)P2 against TRPM3 is shown in Figure 1. As a matter of fact, because of the extreme flexibility of the lipid tail, the docking algorithm failed in generating binding poses for the full-length PI(4,5)P2 lipid. Instead, starting from the PI(4,5)P2 headgroup, a series of truncated versions of a growing lipid were docked successfully against the binding site on TRPM3 until a maximum tail length was reached (our truncated lipid is similar to the synthetic diC8 PI(4,5)P2 molecule, which is experimentally functional in activating TRPM3 (21)). For simplicity, in this work, the PI(4,5)P2 lipid with truncated tails, which was modeled in complex with TRPM3, is referred to as PI(4,5)P2. Note that in for TRPM8, the PI(4,5)P2 molecule was modeled as a full-length lipid. As for the molecular docking, we used the same protocol implemented for TRPM8. Related figures were generated using the Visual Molecular Dynamics (VMD) molecular visualization program (63) (http://www.ks.uiuc.edu/).
A number of structural alignments were performed to compare TRPM3 and TRPM8 structures, including models (TRPM3 and TRPM8) and experimental structures (TRPM8). Superposition of the atomic coordinates was all performed based on sequence alignments (using the algorithm Needelman–Wunsch with BLOSUM-62 matrix). Alignments were generated using the Match Maker tool in UCSF Chimera (64), version 1.15, and analyzed in VMD. A number of structural alignments were performed, described as follows. (1) The model of TRPM8 in complex with (full length) PI(4,5)P2 and that of TRPM3 in complex with (truncated) PI(4,5)P2 was aligned. (2) The TRPM8/PI(4,5)P2 model and the experimentally determined structure of TRPM8 in complex with the menthol analog WS-12 and PI(4,5)P2 (PDB ID: 6NR2), and the complex of TRPM8 with icilin (PDB ID: 6NR3), PI(4,5)P2, and calcium (28). Pairwise backbone RMSD values were calculated for two separate selections (Table 1), including amino acid ranges facing the lipid binding sites, using the VMD RMSD Trajectory Tool. Before RMSD was calculated, structures were aligned on each selection. The first selection (sel-1 in Table 1) included residues 670 to 685 (on pre-S1), residues 724 to 735 (on pre-S1), and residues 851 to 865 (on linker). The second selection (Sel-2 in Table 1) included residues 670 to 685 (on pre-S1), residues 724 to 735 (on pre-S1), residues 851 to 865 (on linker), and residues 997 to 1009 (on TRP domain). (3) The following structures were aligned to the TRPM3–PI(4,5)P2 model: the experimental structure of TRPM4 (29) and TRPM3 models from AlphaFold (DeepMind, EMBL-EBI) (31, 32). At the time of writing, four AlphaFold models were available of TRPM3, each from a different organism (UniProt sequence ID: Q9HCF6 [human; Fig. S1], J9S314 [M. musculus], F1QYX6 [Danio rerio], and F1LN45 [Rattus norvegicus]). All four structures were superimposed (not shown), revealing striking structural similarities. All figures related to (1) to (3) were generated using VMD.
Changes in the binding affinity (or Gibbs free energy of binding, ΔΔG in kcal/mol) of PI(4,5)P2 to TRPM3 were calculated upon mutating key binding residues in the putative PI(4,5)P2 binding site. These residues were also mutated experimentally. To do so, a physics-based scoring was employed (65), previously used with systems similar to the one included in this study (25, 66). Essentially, residue mutations and ΔΔG calculations were performed on the TRPM3 model bound to truncated PI(4,5)P2 molecules as generated from molecular docking, that is, the native structural complexes or WT. A total of three binding modes were used as native configurations, including the TRPM3 model bound to the truncated PI(4,5)P2 presented in this study, and two additional poses with different protonation states of the phospholipid headgroup and even shorter tails. Then, the “Residue-Scanning and Mutation” tool from BioLuminate (65) (Schrödinger, LLC, 2018) was used to perform calculations upon mutating the native system, as described in Table S2. For each of the three WT proteins, six additional mutants were generated, reaching a total of 21 systems. For each system, and for both the WT and the mutant, an MM/GBSA refinement of the bound and unbound states was performed using Prime (Schrodinger, LLC, 2018), via the VSGB 2.0 implicit continuum solvation model (43). In VSGB 2.0, the solvation free energy is approximated with an optimized model based on the surface generalized Born method (44) and the variable dielectric treatment of polarization (45) for protein residues. The latter incorporates the polarization effects by changing the value of the internal dielectric constant (from 1.0 to 4.0) (43). No implicit membrane model (i.e., a low-dielectric slab) was used, and therefore, the results should be regarded as an approximation to the electrostatic energy. The structural complexes were refined by side-chain prediction with backbone sampling/minimization of the mutated residue, before a minimization in the region around the mutation site was performed to relax and optimize the side-chain interactions with the lipid. Systems were prepared for the calculations using the Protein Preparation Wizard (58). A thermodynamic cycle was then used to calculate the change in the binding affinity, , of a protein upon single amino acid mutation, as represented below: The change in the binding affinity (the net free energy difference) was calculated by addressing the free energy changes in vertical lines, easier to simulate than the experimental observables (horizontal lines). A positive value of indicates that the WT binds better than the mutant. Affinity changes were plotted using Microsoft Excel (https://www.microsoft.com/). Related figures were generated using VMD.
All data are contained in the article and supporting information. The structural model of TRPM3 in complex with a PI(4,5)P2 phospholipid with short tails is also available as a supporting information file. The authors request that any published work derived from the use of such data include a reference to this publication.
This article contains supporting information.
The authors declare that they have no conflicts of interest with the contents of this article. |
|
PMC9647549 | Jialu Tu,Haiyang Zhang,Ting Yang,Yun Liu,Solomon Kibreab,Yunpeng Zhang,Liangcai Gao,Robb E. Moses,Bert W. O'Malley,Jianru Xiao,Xiaotao Li | Aging-associated REGγ proteasome decline predisposes to tauopathy | 07-10-2022 | aging,C/EBPβ,protein degradation,tau,REGγ,tauopathy,Aβ, amyloid-beta,AD, Alzheimer’s disease,cDNA, complementary DNA,ChIP, chromatin immunoprecipitation,HA, hemagglutinin,KI, knock-in allele,NFTs, neuron fibrillary tangles,NOI, novel object index,NOR, novel object recognition,OA, okadaic acid,qRT-PCR, quantitative real-time PCR,UPS, ubiquitin-proteasome system | The REGγ-20S proteasome is an ubiquitin- and ATP-independent degradation system, targeting selective substrates, possibly helping to regulate aging. The studies we report here demonstrate that aging-associated REGγ decline predisposes to decreasing tau turnover, as in a tauopathy. The REGγ proteasome promotes degradation of human and mouse tau, notably phosphorylated tau and toxic tau oligomers that shuttle between the cytoplasm and nuclei. REGγ-mediated proteasomal degradation of tau was validated in 3- to 12-month-old REGγ KO mice, REGγ KO;PS19 mice, and PS19 mice with forebrain conditional neuron-specific overexpression of REGγ (REGγ OE) and behavioral abnormalities. Coupled with tau accumulation, we found with REGγ-deficiency, neuron loss, dendrite reduction, tau filament accumulation, and microglial activation are much more prominent in the REGγ KO;PS19 than the PS19 model. Moreover, we observed that the degenerative neuronal lesions and aberrant behaviors were alleviated in REGγ OE;PS19 mice. Memory and other behavior analysis substantiate the role of REGγ in prevention of tauopathy-like symptoms. In addition, we investigated the potential mechanism underlying aging-related REGγ decline. This study provides valuable insights into the novel regulatory mechanisms and potential therapeutic targets for tau-related neurodegenerative diseases. | Aging-associated REGγ proteasome decline predisposes to tauopathy
The REGγ-20S proteasome is an ubiquitin- and ATP-independent degradation system, targeting selective substrates, possibly helping to regulate aging. The studies we report here demonstrate that aging-associated REGγ decline predisposes to decreasing tau turnover, as in a tauopathy. The REGγ proteasome promotes degradation of human and mouse tau, notably phosphorylated tau and toxic tau oligomers that shuttle between the cytoplasm and nuclei. REGγ-mediated proteasomal degradation of tau was validated in 3- to 12-month-old REGγ KO mice, REGγ KO;PS19 mice, and PS19 mice with forebrain conditional neuron-specific overexpression of REGγ (REGγ OE) and behavioral abnormalities. Coupled with tau accumulation, we found with REGγ-deficiency, neuron loss, dendrite reduction, tau filament accumulation, and microglial activation are much more prominent in the REGγ KO;PS19 than the PS19 model. Moreover, we observed that the degenerative neuronal lesions and aberrant behaviors were alleviated in REGγ OE;PS19 mice. Memory and other behavior analysis substantiate the role of REGγ in prevention of tauopathy-like symptoms. In addition, we investigated the potential mechanism underlying aging-related REGγ decline. This study provides valuable insights into the novel regulatory mechanisms and potential therapeutic targets for tau-related neurodegenerative diseases.
Aberrant accumulation of filamentous tau lesions, which are a characteristic feature of Alzheimer’s disease (AD) and tauopathies, is the most common neuropathological manifestation in several neurodegenerative diseases, such as progressive supranuclear palsy, Pick’s disease, frontotemporal dementia with parkinsonism linked to chromosome 17, and corticobasal degeneration (1). The etiological factors associated with neurodegenerative dementia include vascular, inflammatory, and metabolic factors. The most substantial overall risk factor for neurodegenerative dementia is aging. Aging of a population is associated with an increased incidence of AD, which affects more than 35 million individuals worldwide (2). The pathological features of AD are hyperphosphorylation of tau proteins in neuronal cells (leading to the formation of neuron fibrillary tangles (NFTs) (3)) and amyloid plaques (resulting from extracellular amyloid-beta [Aβ] deposition (4)). NFTs are associated with neuronal death and cognitive impairment (5). Although tau is mainly an intraneuronal protein, autopsy analysis of the brain of patients with AD has revealed that the pathological impact of NFTs is stratified (6). The formation of NFTs is initiated at the transentorhinal cortex and subsequently develops in the synaptic areas of the brain, such as the hippocampus, or the new cortex (7). Previous studies have reported the aging-dependent roles of nuclear tau in neurodegeneration (8, 9). Recent studies indicated that tau, not Aβ, may be the key etiological factor for the symptoms of AD (10) and that tau deposits are a biomarker for monitoring AD (11), as well as tauopathies. Although many Aβ-targeted drugs in AD treatment have failed to show efficacy, the FDA recently approved one of the anti-Aβ antibodies to remove amyloid plaque from AD brains (12). For the other major AD pathological lesion, NFTs consisting of phosphorylated tau (p-tau) (13), and related research of tau-targeted treatment aiming to clear NFTs in AD brains also has appeared to be promising. It remains a formidable task to ensure that these targeted therapies have a demonstrated clinical efficacy. The proteasome is reported to play an indispensable role in maintaining protein homeostasis and mediating neuronal apoptosis and synaptic plasticity (14, 15, 16). The accumulation of misfolded tau proteins, such as phosphorylated tau and NFTs, can impair the function of the 26S proteasome complex (3, 17), which increases the susceptibility of neurons to degeneration (18). REGγ codes for REGγ (also known as PA28γ, PSME3, Ki antigen, and the 11S family proteasome activator) (19), a noncanonical proteasome activator mediating ubiquitin-independent and ATP-independent protein degradation (20); it has been reported to decrease polyglutamine-expanded androgen receptor aggregation and consequently alleviate motor muscle atrophy and spinal and bulbar muscular atrophy (21). The expression of REGγ is upregulated in the neurons of human and mouse brains (22). Recent single cell RNA-seq and proteomic analyses have indicated that the expression of REGγ is markedly downregulated in aged individuals and patients with AD (23, 24). Consistent with these reports, bioinformatics analysis revealed that the expression of REGγ is downregulated in multiple tissues of patients with AD (25, 26, 27). The levels of REGγ were inversely correlated with those of tau. Thus, a panel of mutant REGγ derivative mice combined with the P301S Tg tau (PS19) model (3) was generated to elucidate the role of REGγ in AD. The PS19 model mouse has a transgenic human tau gene with a P to S change at position 301, a mutation found in human tauopathy. The findings of this study demonstrate that REGγ downregulation accelerates tau deposition and its effects, whereas the overexpression of REGγ ameliorates tau lesions in PS19 mice.
Previously, we had reported that the depletion of REGγ in mice results in premature aging (28). Thus, in the present study, we aimed to evaluate REGγ expression patterns during aging and in age-related disorders. To determine REGγ profiles during physiological aging in mice, the REGγ expression levels were determined in mice aged 2 to 24 months (Figs. 1A and S1A). The levels of REGγ progressively decreased starting at the age of 5 months. Similar age-dependent reduction of REGγ was observed in the tauopathy model PS19. At 2 months of age, the expression of REGγ in PS19 mice was downregulated compared with that in the WT control (Figs. 1A and S1A). Next, bioinformatics analysis was performed to determine REGγ expression in the publicly available Gene Expression Omnibus (GEO) datasets of the National Center for Biotechnology Information database. The dataset comprised the gene expression profiles of postmortem AD brains. In particular, GSE1297 (26) (dataset 1) comprised 31 independent microarray data of nine healthy controls and 22 patients with AD, in which seven were severe cases. GSE159699 (27) (dataset 2) comprised the lateral temporal lobe RNA-seq analysis results of 12 patients with AD and 10 aged controls. Compared with those in the healthy controls, the REGγ mRNA levels were significantly downregulated in patients with AD. In particular, the REGγ mRNA levels were markedly downregulated in the hippocampal interneurons of patients with AD (Fig. 1B), with similar trends of REGγ decline in brain cortex and hippocampus in AD patients (Fig. S1B). To validate these findings, postmortem AD samples obtained from the Association of Human Brain Bank of China were subjected to immunohistochemical (IHC) analysis (29, 30). The expression of REGγ in the hippocampus of all five AD cases was markedly lower than that in the hippocampus of age-matched healthy controls (Fig. 1C). The staining of the same regions with anti-p-tau (AT8) antibodies revealed that the tau levels in AD specimens were significantly higher than those in healthy control specimens (Fig. 1D). This indicated that the levels of REGγ were inversely correlated with those of tau in the AD brain lesions. These results suggest a potential role for REGγ in age-related dementia.
To determine if REGγ regulates the levels of tau and p-tau, REGγ knockdown SH-SY5Y cell lines were established by transfecting cells with shRNA against REGγ (sh-REGγ or shR). Control shRNA (shN)–transfected cells were used as controls (31). The levels of tau (total tau/p-tau levels) were markedly higher in shR-transfected SH-SY5Y cells (Figs. 2A, left panel, Fig. S1D) and si-REGγ-transfected HT22 cells (Fig. 2A, right panel and Fig. S1D). However, transfection with shR and si-REGγ did not markedly affect the total MAPT or mapt mRNA levels, respectively (Fig. S1, F and G). Next, embryonic primary neuronal cells were isolated from the hippocampus of four different genotypes of mice for in vitro studies. The expression levels of human MAPT (in PS19 or P301S Tg, a mutant human Tau-overproducing mouse line) and mouse mapt (total tau/p-tau) were higher in the REGγ knockdown and REGγ KO;PS19 neurons (Figs. 2B and S1E), and a similar tendency displayed in mice brain hippocampus tissues (Fig. S2A). The degradation dynamics of tau and p-tau in shR-transfected and shN-transfected SH-SY5Y cells were analyzed in the presence of cycloheximide, a protein synthesis inhibitor. The decay of total tau (HT7) and p-tau (p-tau T231 and p-tau S396) proteins in shR-transfected cells was markedly slower than that in the shN-transfected cells (Figs. 2C and S2F). This suggested that REGγ regulates the stability of tau and p-tau in these cells. Since identified REGγ substrate proteins must interact with the REGγ activator, the physical interaction between REGγ and tau proteins was examined using reciprocal coimmunoprecipitation assays with anti-REGγ, anti-total tau, or antihemagglutinin (HA)/GFP antibodies. Endogenous and exogenously expressed REGγ interacted with tau in cultured cells or hippocampus tissues of PS19 mice (Figs. 2D and S2, B and C). To determine the direct role of the REGγ-20S system in the degradation of tau and a mimic-phosphorylated tau, cell-free proteolysis was performed with purified proteins in vitro. Translated tau and tauS396E (a phosphorylation-mimetic mutant) were not significantly degraded upon incubation with 20S proteasome or REGγ (Fig. 2E; lanes 2 and 3). However, the combination of REGγ and 20S proteasome effectively degraded tau and tauS396E (Fig. 2E; lane 4). Soluble oligomers of tau protein are reportedly more toxic than the p-tau aggregates (32). We wondered if REGγ may also regulate the levels of soluble tau oligomers. Okadaic acid (OA), an efficient selective inhibitor of protein phosphatase 2A (PP2A) and protein phosphatase type 1 (PP1), was used to allow the accumulation of tau oligomers that can be recognized by a specific antibody (anti-Tau, T22) (33). The levels of soluble tau oligomers in OA-treated shR-transfected SH-SY5Y cells were 10% higher than those in OA-treated shN-transfected SH-SY5Y cells. This indicated that REGγ degrades soluble tau oligomers (Fig. 2F). These findings indicate that REGγ is directly involved in the degradation of multiple tau species in cells, as well as in a cell-free system.
To elucidate the mechanism underlying REGγ proteasome-mediated turnover of phosphorylated tau, phosphorylation-mimetic (S396E and T231E) and phosphorylation-defective (S396A and T231A) mutant tau constructs were generated. Kinetic studies performed in the presence of cycloheximide revealed that the degradation of phosphorylation-mimetic tau mutants (S396E and T231E) in WT 293T cells was markedly faster than that in REGγ KO cells (34) (Figs. 3A and S2G). In contrast, the decay rates of phosphorylation-defective tau mutants (S396A and T231A) were similar in WT and REGγ KO 293T cells (Figs. 3B and S2H). This suggested that the REGγ proteasome primarily targets phosphorylated tau for degradation. REGγ, a nuclear protein, is thought to mediate the degradation of nuclear proteins, as well as the degradation of cytoplasmic proteins that shuttle between cytosol and nuclei (35). Next, we investigated the effect of phosphorylation on the cellular distribution of tau in SH-SY5Y cells. OA treatment promoted the nuclear localization of tau in more than 90% of the cells (Fig. 3C), suggesting a role of phosphorylation in the nuclear translocation of tau. To test this, WT tau or phosphorylation-mimetic/defective tau (with a single mutation) constructs were generated and exogenously expressed in WT or REGγ-deficient cells. The nuclear translocation of WT human tau in REGγ-deficient cells was approximately 50% more than that in WT controls (Fig. 3D). The expression of phosphorylation-mimetic tau enhanced the nuclear translocation of tau by approximately 90%. In contrast, transfection with phosphorylation-defective tau did not affect its nuclear translocation (Figs. 3, E and F and S2E). Consistent with the immunostaining results, the expression of WT tau or phosphorylation-mimetic tau (but not that of phosphorylation-defective tau) in REGγ-deficient cells was upregulated compared with that in control cells (Fig. S2D). These findings suggest the nuclear translocation of phosphorylated tau may explain the reason for REGγ primarily mediating the degradation of nuclear tau.
To investigate REGγ-mediated regulation of tau in vivo, transgenic mice with forebrain neuron-specific overexpression of REGγ were generated after crossing REGγ knock-in allele (KI) mice with Camk2α-cre mice (Fig. S3A). REGγ KI mice were obtained without any changes in the brain REGγ level compared to the REGγ WT; thus, either mouse group could be used as REGγ normal controls. Thus REGγ KI and REGγ WT mice were used as control mice for REGγ KO and REGγ OE mice. Similarly, the mice in the PS19 group, REGγ KI;PS19, and REGγ WT;PS19 were used as the control group against REGγ KO;PS19 and REGγ OE;PS19 mice. REGγ KI and REGγ WT mice were crossed with PS19 mice to generate Control;PS19 mice with the same REGγ levels. Quantitative real-time PCR (qRT-PCR) analysis revealed that the REGγ levels were highest in the hippocampus and cortex of Camk2α-cre mice with homozygous REGγ KI (Homo-REGγ OE). However, the REGγ level in the cerebellum of Camk2α-cre mice with heterozygous REGγ KI (Hetero-REGγ OE) or Homo-REGγ OE mice were not upregulated compared with that in the cerebellum of the controls (Fig. S3, B–D). This was consistent with the expectation that Camk2α-cre drives REGγ expression in forebrain neurons. To validate the conditional expression of the Flag-REGγ KI allele, homogenized forebrain tissues (REGγ OE mice forebrain tissues) were immunoprecipitated with anti-Flag antibodies. Mass spectrum analysis revealed the expression of exogenous REGγ alleles in the mouse brain (Fig. S3E). IHC analysis with anti-Flag (left and middle panels) or anti-REGγ (right panel) antibodies revealed the differential expression of REGγ in the hippocampus of REGγ-overexpressing and normal control mice (Fig. S3F). Next, the REGγ KO or REGγ OE mice were crossed with PS19 mice to generate REGγ KO;PS19 or REGγ OE;PS19 mice, respectively. To examine the effects of REGγ levels on various tau species in vivo, the hippocampal tissues of mice aged 8 and 10 months from the six different genotypes were analyzed by Western blotting analysis with p-tau–specific antibodies. The total tau (HT7) and p-tau (pS396 and pT231) levels in both REGγ KO and REGγ KO;PS19 (human tau species with slower migration) mice were significantly higher compared with those in Control or Control;PS19 (Fig. 4, A and D). Increased p-S202/T205 (AT8) staining intensity was only observed in REGγ KO;PS19 mice (Fig. 4A) but not in Control;PS19 mice. This suggested that REGγ depletion promotes tau hyperphosphorylation in PS19 mice. In contrast, the hippocampus of REGγ OE;PS19 mice exhibited significantly lower expression of total tau (HT7) and p-tau (detected using p-T212/S214 [AT100]) and AT8 than that of Control;PS19 mice (Fig. 4, B–D). Tau is reported to accumulate with aging in the neurons of the hippocampus and is concomitantly downregulated in tau immune-reactive CA3 mossy fibers (3). The results of this study were consistent with this observation (see arrow heads in Fig. 4E). Comparison of the tau staining revealed the levels of nuclear tau in REGγ KO;PS19 and REGγ OE;PS19 mice (Fig. 4G). Our findings of higher levels in the KO were consistent with the observation in cultured cells treated with OA (Fig. 3C) or expression of phosphorylation-mimetic tau (Fig. 3F). IHC analysis revealed that p-tau stained with T231 (AT180) or AT8 antibodies was markedly upregulated in REGγ KO;PS19 but downregulated in REGγ OE;PS19 mice (Figs. 4F and S3G). These findings were consistent with those of Western blotting analysis (Fig. 4, A–D) and demonstrate the effects of REGγ on tau protein levels.
To evaluate the effect of REGγ levels on brain atrophy and cytoarchitecture in PS19 models, computer-assisted image analysis of brain size and the number of neurons in CA1 regions in mice belonging to the six different genotypes was performed. The size of the whole brain and hippocampus was not remarkably different among Control, REGγ KO, and REGγ OE mice aged 10 months. In contrast, Control;PS19 mice exhibited marked brain atrophy, while ventricular dilation was observed in age-matched REGγ KO;PS19 mice. These pathological changes were alleviated in REGγ OE;PS19 mice (Fig. 5A). Nissl staining analysis of the neuron layer thickness in the hippocampus revealed an increased neuron degeneration in PS19 (Control;PS19) mice aged 10 months that was further exacerbated (thinner) in REGγ KO;PS19 mice but significantly alleviated in REGγ OE;PS19 mice (Fig. 5A, lower panel). The density of neurons in PS19 and REGγ KO mice appeared to be less than that in Control and REGγ OE mice. To determine the neuronal loss in the CA1 regions, the brains of the different mouse genotypes were stained with anti-NeuN antibodies. Quantitative analysis of CA1 regions revealed 69% and 44% of control neurons in REGγ KO and REGγ KO;PS19 mice, respectively (Fig. 5B). Mice with compound mutations in REGγ and tau (REGγ KO;PS19) exhibited more than 50% loss in CA1 neurons, which was significantly alleviated in REGγ OE;PS19 mice (Fig. 5B). We found similar changes in dentate gyrus regions of corresponding mice (Fig. S3, H and I). Gliosis is associated with tau lesions and/or neuronal loss in tauopathies (36, 37). Hence, mouse brains from different genotypes were stained using anti-glial fibrillary acidic protein (GFAP) antibodies. GFAP signals were observed throughout the whole brain of PS19 mice. We also found increased GFAP staining in the white and gray matter of the hippocampus and other brain regions in REGγ KO;PS19 (Fig. 5C). The levels of GFAP were significantly attenuated in REGγ OE;PS19 (Fig. 5C). This suggested a reduction in astrogliosis, which may be due to attenuated tau lesions and neuron loss. Consistent with these observations, Gallyas–Braak silver staining revealed that the number of NFTs (red arrow heads) in REGγ KO;PS19 mice was more than that in Control;PS19 or REGγ OE;PS19 mice (Fig. 5D). Loss of synapse formation is reported to be an early marker in the PS19 tauopathy model (3). Golgi staining was performed to analyze the dendritic spines in the cortex. The results of all animals in each genotype were averaged. Compared to Control mice, the number of dendritic spines was lower in REGγ KO mice and further decreased in REGγ KO;PS19 mice (Fig. 5E). The dendrite abnormality in REGγ KO;PS19 mice was alleviated with increased spine density in REGγ OE;PS19 mice (Fig. 5E). These results suggest that loss of REGγ potentiates neurodegenerative phenotypes in PS19, and these changes can be alleviated by restoring REGγ function.
Impaired learning and memory are the hallmarks of human tauopathy. Previously, we had reported the effect of REGγ levels on the hippocampus-dependent spatial memory using the Morris water maze test (38). Mice belonging to six different genotypes without significant differences in swimming speed or motor activity were screened out (Fig. S4, A and B). The learning of 6-month-old and 9-month-old REGγ OE;PS19 mice was faster than that of Control;PS19 and REGγ KO;PS19 mice (Fig. S4C). The percentage of time spent in the target quadrant and the number of times to the hidden platform by REGγ OE mice were significantly higher than those by Control and REGγ KO littermates. Similarly, the percentage of time spent in the target quadrant and the number of visits to the platform by REGγ OE;PS19 mice were higher than those in Control;PS19 and REGγ KO;PS19 mice (Figs. 6A and S4E). During the reversal probe trial in which the target platform was switched to the opposite quadrant, the learning of latency to reversal platform of REGγ OE;PS19 mice was faster than that of Control;PS19 and REGγ KO;PS19 mice (Fig. S4D). Moreover, Control;PS19 and REGγ KO mice spent an increased amount of time in the primary platform target quadrant and decreased amount of time in the reversal quadrant (Fig. 6B). Spatial learning ability was examined using the radial eight-arm maze task (39). Overexpression of REGγ decreased the number of errors in both REGγ-deficient and PS19 mice, including REGγ KO;PS19 mice (Fig. 6C). These results indicate that REGγ is crucial for hippocampus-dependent spatial memory. To further examine the effect of REGγ dysfunction on cognitive and noncognitive (such as anxiety and motivation) impairments in Control;PS19 mice, novel object recognition (NOR) (to evaluate the hippocampus-dependent short-term memory) and elevated plus maze (EPM) tests were performed. In mice exhibiting a similar discrimination index for objects A and B, the novel object index (NOI) was measured before and after switching object B to a different object C. The NOIs of REGγ OE;PS19 mice were significantly higher than those of REGγ KO;PS19 littermates (Fig. S4F). EPM was used to investigate anxiety based on the natural spontaneous exploratory behavior of mice in novel environments, as well as on their natural aversion for elevated and open areas, and the tau mutant transgenic mice spent more time in the open arms, indicating that their anxiety might be lower (40). The anxiety levels in REGγ OE;PS19 mice were significantly higher than those in Control;PS19 and REGγ KO;PS19 mice, indicating that REGγ activity can prevent anxiety behavior in the mouse models (Fig. 6D). In addition to the beneficial effect of REGγ overexpression on neurodegenerative phenotypes in PS19 mice (REGγ OE;PS19 mice), the life span of REGγ OE;PS19 mice was significantly longer than that of Control;PS19 and REGγ KO;PS19 mice (Figs. 6E and S4G). These results demonstrate that increased REGγ activity alleviates tauopathy-induced cognitive deficits and promotes prolonged survival of mice.
To determine if the reduced REGγ expression in aged and AD/tauopathy brains (Fig. 1) resulted from dysregulation or loss of neurons, transcriptional regulation of REGγ in neuronal cells was examined in vivo and in vitro. CCAAT enhancer-binding protein-beta (C/EBPβ), a transcription factor that is activated in response to inflammation regulates a panel of factors, such as δ-secretase and apolipoprotein E ε4 (APOE4) (41, 42) C/EBPβ, is upregulated in the aged brain (43) and is reportedly a factor that induces cognition defects in mice (44). Transforming growth factor beta receptor (TGFβR), which is a transcription target of C/EBPβ (45), mediates a signaling pathway to repress transcription of REGγ (46). Compared with that in the hippocampal tissues of 24-month-old mice, the REGγ mRNA levels were downregulated and the Cebpb and Tgfbr2 mRNA levels were upregulated in the hippocampal tissues of 3-month-old mice (Fig. 7A). Furthermore, transfection of CEBPB into SH-SY5Y cells upregulated the expression of TGFBR2 and significantly downregulated the REGγ levels (Figs. 7B and S1C). These results suggest that age-dependent reduction of REGγ may result from dysregulation of the C/EBPβ signaling pathway. Based on the findings of this study, we propose a model for the role of the proteasome activator REGγ in the regulation of tau homeostasis. The REGγ-20S system degrades tau species, including tau oligomers. Genetic ablation of REGγ promotes tauopathies in PS19 models, whereas conditional activation of REGγ expression in the forebrain neurons rescues tau lesion and aging-associated neurodegenerative phenotypes. With age, C/EBPβ signaling activation leads to a decline in the levels of REGγ in association with inflammation. Concomitantly, the loss of REGγ promotes the nuclear translocation of tau, which may promote pathological function of tau other than aggregates. Overexpression of CEBPB in human neuroblastoma cells (SH-SY5Y cell) downregulated the REGγ mRNA levels and upregulated the TGFBR2 mRNA levels (Figs. 7B and S1C). Moreover, binding of C/EBPβ to the TGFβR2 locus has been reported in the chromatin immunoprecipitation (ChIP) sequencing database in human A549 cell line. ChIP assays performed in the SH-SY5Y cell line indicated that C/EBPβ could be recruited to the TGFβR2 promoter in a neuronal cell line (Fig. 7C). C/EBPβ-induced activation of TGFβ signaling via promotion of TGFβR2 was evidenced by enriched Smad3 on REGγ promoter, but not in regions further upstream, by ChIP analyses in SH-SY5Y cell line (Fig. 7D). Compared with those in the brain hippocampus of 3-month-old mice, the Cebpb and Tgfbr2 levels in 24-month-old mice were upregulated and the REGγ levels were markedly downregulated in the brain (Fig. 7A). Therefore, age-related REGγ reduction appears to be regulated by C/EBPβ through the TGFβ signaling pathway. The findings of this study may be clinically relevant for the development of new therapeutic strategies for neurodegenerative diseases, such as tauopathies and AD.
This study demonstrated that REGγ plays a critical role in the regulation of hippocampus-dependent learning and memory in AD-like syndromes by directly targeting tau and p-tau for proteasome-mediated degradation. REGγ deficiency markedly upregulated the levels of phosphorylated tau in the nuclei, promoted the accumulation of toxic tau oligomers, and consequently potentiated neurodegenerative tauopathy in mouse models. This study presents proof-of-principle evidence for neuron-specific REGγ expression-mediated mitigation of the progression of tauopathy or AD-like symptoms. Mechanistic studies led to a proposed link between aging-associated REGγ downregulation and tau-related neurodegeneration (Fig. 7E). We found that aging and aging-associated degenerative dementia were associated with downregulated REGγ expression. This is consistent with the results of a previous study, which reported that REGγ deficiency promotes premature aging in mice (28). The findings of this study are consistent with those of a previous study (47). Additionally, this study demonstrates that the mRNA and protein levels of REGγ were upregulated in the pyramidal neurons of healthy brain regions, including the hippocampus (Allen Brain Institute https://mouse.brain-map.org/ and the Human Protein Atlas Institute http://www.proteinatlas.org/). Previously, we had proposed a mechanism through which REGγ is downregulated by C/EBPβ (43) via the TGFβ (46) signaling pathway. However, we do not exclude the possibility of additional factors contributing to aging-associated REGγ reduction. To the best of our knowledge, this is the first study to report the degradation of tau and p-tau proteins by the ubiquitin-independent REGγ-proteasome system. Various tau clearance pathways have been previously reported. The major intracellular degradation processes are ubiquitin-proteasome system (UPS) and autophagy (48). These tau degradation pathways can act on different forms of tau protein. Excessive soluble neurotoxic tau proteins can be degraded through the UPS (49), chaperon-mediated autophagy (50), and endosomal microautophagy (51). Meanwhile, the intraneuronal insoluble tau is degraded via macroautophagy (52). Based on our previous results, all the substrate proteins identified to be targeted by the REGγ-proteasome system are also regulated by UPS. In most cases, UPS mediates signal-mediated acute degradation of protein substrates, whereas the REGγ-proteasome system primarily maintains the steady state levels of these proteins. We believe the ubiquitin-dependent and ubiquitin-independent regulation of tau will be orchestrated in similar fashion under normal conditions. However, both UPS and autophagy pathways are impaired in several neurodegenerative diseases (53, 54). This suggests the importance of the REGγ pathway in the maintenance of the homeostasis of key cellular proteins including tau proteins. The identification of REGγ-mediated tau degradation provides additional therapeutic targets for aging-associated neurodegeneration. For more than 3 decades, Tau proteins have been reported to localize to the neuronal and non-neuronal cell cytoplasm, as well as to the nucleus, (8). However, most studies have focused on the role of tau in the physiological and pathological processes in the context of the microtubules. Recent studies considered nuclear tau as a molecular marker of cell aging and aging-associated diseases, such as AD (8). Nuclear tau is reportedly indispensable for cellular responses against cellular injury and DNA damage (8, 55). Additionally, nuclear tau can organize and protect the chromatin during cellular aging (8, 56). The functional nucleolar tau is mostly dephosphorylated. Upon phosphorylation, tau dissociates from the DNA (55). The absence of functional tau due to mutation (such as P301L and P301S) might impair the genome-protective functions of tau and render the cells susceptible to chromosomal instability (55). REGγ deficiency or dysfunction also promotes genome instability (57). The present study demonstrated that the expression of tau is correlated with that of REGγ. REGγ depletion promoted the accumulation of phosphorylated tau in the nuclei, which suggested the correlation between REGγ-proteasome function and nuclear tau regulation. Future studies should focus on the roles of REGγ and nuclear tau in inducing genome instability during the pathogenesis of neurodegenerative diseases. Previously, we had demonstrated that the accumulation of GSK3β contributes to the development of brain disorders in aged REGγ KO mice (47). GSK3β is an important kinase involved in the hyperphosphorylation of tau and the pathogenesis of aging-associated dementia (58, 59). Therefore, the loss of REGγ function may regulate the pathogenesis of neurodegenerative diseases at multiple levels. REGγ overexpression significantly mitigated the progression of neurodegenerative disorders (including AD-like cognitive impairments), loss of neurons and dendritic spines, formation of NFTs, and reduction of life span in mice. Interestingly, REGγ was reported to play an important role in innate immune responses and inhibits the overactivation of immunoproteasome and consequential development of autoimmune diseases (60, 61). Our observation of microglial activation is evidenced by increased GFAP staining in the hippocampus and other brain regions in REGγ KO;PS19 in mice, suggesting a potential role of REGγ in the regulation of immune responses in neural system. Detailed molecular mechanisms by which REGγ deficiency enhance microglial activation need further analysis. In summary, the findings of this study demonstrate that REGγ downregulation during aging or in age-related brain disorders is associated with predisposition to tauopathies and AD. REGγ-mediated proteasomal degradation of tau, especially phosphorylated tau, is a novel mechanism for the regulation of tau homeostasis. This may help to identify novel roles of nuclear tau in addition to its role as a microtubule-associated protein. Strategies to achieve REGγ gain of function may aid in the development of novel therapies for tau-related neurodegenerative diseases.
All animal experiments were conducted according to the guidelines of the Institutional Animal Care and Use Committee of East China Normal University and human sample experiments were conducted according to the guidelines of the University Committee on Human Research Protection with the ethical approval number: HR 016-2021. The animal ethical committee approval number is m20200303. REGγ KO mice (C57BL/6J background) were generated as reported previously (47). Mice with cre transgenes (Camk2α-cre) and conditional REGγ alleles with the R26-stop-FLAG reporter (conditional REGγ KI) were maintained under the same conditions described previously. Camk2α-cre mice and conditional REGγ KI mice were hybridized over 10 generations to obtain the stable genotype. REGγ OE;PS19 mice were originally from JAX Laboratory with Prnp-MAPT∗P301S mutation on a mixed B6; C3-Tg background. PS19 mice were crossbred with REGγ KO, Camk2α-cre, and conditional REGγ KI mice over 10 generations to obtain the stable C57BL/6J background offspring Control;PS19, REGγ KO;PS19, and REGγ OE;PS19, respectively. Male C57BL/6J mice aged 3 to 24 months were used unless otherwise described. All animals were bred in the animal room under the following conditions: temperature, 20 to 25 °C; humidity, 40% to 70%; circadian cycle, 12 h light/dark cycle; food and water supply, ad libitum.
The PSME3 expression data were obtained from the GEO database. The GSE159699 dataset included the RNA-seq data of postmortem lateral temporal lobe of patients with AD (n = 12), aged healthy control (aged, n = 10), and young healthy control (young, n = 8). Gene expression was normalized to FPKM. PSME3 expression in the old (n = 10) and AD (n = 12) datasets was comparatively analyzed. Additionally, the gene expression data GSE1297, which is a microarray data of hippocampal gene expression in healthy control and patients with AD exhibiting varying severity, were downloaded. PSME3 expression in healthy control (n = 9) and severe AD (n = 7) cases were analyzed. RNA-seq and microarray data were separately analyzed using GEO query and Limma packages in R http://www.r-project.org/ as described (26).
HEK 293T (WT and REGγ knockout using TALENs), SH-SY5Y (shN-transfected and shR-transfected), and HT22 (control and si-REGγ-transfected) were used in this study (31, 47, 62). si-REGγ RNA sequences were shown in the Table S1. All cells originally obtained from ATCC were cultured in Dulbecco’s modified Eagle’s medium (DMEM) or DMEM/F-12 (1:1) supplemented with 15% fetal bovine serum (HyClone), 100 IU/ml penicillin, and 100 mg/ml streptomycin (Thermo Fisher Scientific) in a humidified incubator at 37 °C and 5% CO2. Silencing RNA sequences were listed in the Table S3.
Hippocampus and auditory cortex were dissected and subjected to SDS-PAGE. The resolved proteins were transferred to a nitrocellulose membrane and the protein signals were detected using fluorescent secondary antibodies in the Image Studio system, following routine protocols. Antibodies were shown in the Table S2.
Mice brain tissues were perfused with ice-cold PBS (1×) and 4% paraformaldehyde and fixed with 4% paraformaldehyde for 72 h at 4 °C. To terminate fixation, the brain tissues were incubated in a solution containing 4% acrylamide, 1 M glycine, and 0.1% Triton-X 100 in 1× PBS for 48 h at room temperature. The tissues were washed with 1× PBS and sectioned into 10 μm thick sections using a freezing microtome (Leica CM1950) in 1× PBS. The detailed staining process has been described elsewhere (47). The images were captured using Tissue Gnostics Tissue FAXS Plus ST (ZEISS) and THUNDER Image System (Leica) microscope.
Nissl staining was performed using the kit from Beyotime Biotechnology (C0117), following the manufacturer’s instructions. The sections adjacent to the stained area were selected to measure the size of the whole brain and the number of neurons using the software Unbiased Stereology Tissue FAX Plus ST (Tissue Gnostics). Golgi staining was performed with the FD rapid Golgi stain kit (FD Neuro Technologies, Inc), following a previously published protocol (63). Mice were perfused following routine protocols. The sections were also subjected to a modified Gallyas–Braak staining (64). The images were captured using Tissue Gnostics Tissue FAXS Plus ST and fluorescence microscope (Olympus DP74).
REGγ heptamers and 20S core proteins were purified as described previously (20). The target protein tau and tau (S396E) were translated using a translation kit with an appreciate reaction system (Promega), following the manufacturer’s instructions. Degradation reaction conditions have been described elsewhere (20). The reaction mixture was incubated 30 °C for 3 h. All proteins were detected by Western blotting.
The pcDNA3.1-flag-REGγ and pcDNA3.1-GFP-REGγ constructs previously generated (34) were used in this study. Based on the Homo sapiens tau sequence, pcDNA3.1-Flag-tau, pcDNA3.1-GFP-tau, and PSG5-HA-tau constructs were generated. The mutant constructs PSG5-HA-tau (T231A), PSG5-HA-tau (T231E), PSG5-HA-tau (S396A), and PSG5-HA-tau (S396E) were generated based on the primary construct aforementioned. Primers were listed in Table S1 in the supplementary.
Total RNA was extracted from cells and mouse brain tissues using an RNA extraction reagent (Vazyme). The isolated RNA was reverse transcribed into complementary DNA (cDNA) using the Strand cDNA synthesis kit (Vazyme) in a 30 μl reaction mixture. The cDNA was subjected to qRT-PCR using ChamQ SYBR qPCR Master Mix (High ROX Premixed) (Vazyme) and Quant Studio 3.0 (Thermo Scientific). Each experiment was repeated in triplicates. The primer pairs used for quantitative PCR were listed as shown in Table S1 in the supplementary.
The transfection of 293T cells was performed as described previously. SH-SY5Y cells and the hippocampal tissues of PS19 mice were subjected to immunoprecipitation. Cells or tissues were collected and lysed as previously described (34). The flag-beads and protein A/G with antibodies were used to immune-precipitate the specific proteins. Immunoprecipitates were washed thrice with buffer. The samples were centrifuged and the pellets were suspended in protein loading buffer with SDS and subjected to Western blotting analysis. Antibodies used were listed as shown in Table S2 in the supplementary.
ChIP assay was conducted according to a protocol from the Cold Spring Harbor Laboratory published online at http://cshprotocols.cshlp.org/. Primers and antibody used were listed as shown in Tables S1 and S2 in the supplementary.
Mouse activity in an open field was measured using the TruScan system (Coulbourn Instruments). Briefly, mice were placed in a 38 cm × 27 cm × 27 cm chamber with 50 lux illumination. The free locomotion of the mice for 15 min was tracked using Truscan 2.1. Locomotion was recorded every 5 min.
All experiments were performed in a pool (80 cm in diameter) filled with water at 24 to 26 °C to a depth of 1 to 2 cm over the platform. Mice aged 6 and 9 months were used for the experiment. For training, a submerged platform was placed in the center of a quadrant to enable the animal to determine the location of the platform, which was the only escape from the water. On day 6 (probe trial), the platform was removed, and each mouse was placed into the pool from one point. The route taken by the animal in the target quadrant was monitored for 30 s. All sessions (acquisition phase, probe trials, and reverse phase) were tracked using Ethovision XT14 software package (Noldus IT). Latency that monitors the time to locate the hidden platform under the water and platform crossing that indicates the number of times each mouse tries to swim over the removed platform was quantified. A cued platform was used to exclude the potential impact of motor dysfunction in PS19 mice.
The eight-arm radial maze test enables the identification of mice that exhibit age-related AD progression (39). Before training, these mice had limited to no access to food to ensure that they were motivated to search for food in the maze during the test. The baits were restricted to the food cups. During the first 4 days of training, the food pellets (approximately 45 mg in weight) were placed in the food cups (each eight-arm terminal) and the central octagonal plate. The mice of the same group were allowed to search for food together for 10 min. On day 5 (test day), the pellets in all eight-arm terminal cups were placed in a single food cup. Every test was continued until all eight food pellets had been consumed or until 10 min had elapsed. The number of reference and working memory errors was determined.
To examine the recognition memory of mice, each mouse was allowed to move freely in an arena (27 cm length × 27 cm width × 27 cm height) for 3 days. During the first 3 days, the mice were allowed to adapt to the box for 10 min. On day 4, every mouse was allowed to freely explore the arena with two identical objects for 15 min and rest in the cage for 1 h after exploration. One of the objects was replaced with a new object with a similar material. The mouse was then allowed to freely explore the arena for 5 min. The time spent exploring the novel and familiar objects was recorded. The object was judged to be explored when the mouse touched the object with the nose, mouth, and front paw or when the nose, mouth, and front paw were at a distance of ≤2 cm from the object. The NOR index (NOI) was calculated as follows: NOI = new object exploration time/(new object exploration time + old object exploration time) × 100%. NOIs >50 and ≤50 indicate complete and incomplete new object recognition, respectively (65).
Noncognitive deficits, such as anxiety and motivation are factors that may affect cognitive outcomes. The EPM apparatus comprised a gray poly vinyl chloride ‘+’ maze with two open arms, two enclosed arms, and a central platform linking the arms. The apparatus was raised 40 cm above the floor. Each mouse was placed at the central platform facing an enclosed arm and allowed to freely explore for 5 min. Ethovision XT14 software package (Noldus IT) was used to record the progress of this experiment.
Quantitative data of independent samples were analyzed using GraphPad Prism 8.0 (GraphPad Software Inc) and represented as mean ± SEM. The means (including behavioral and image data) were compared using one-way ANOVA, two-way ANOVA, and two-tailed t test.
All raw bioinformatics analysis data are available in the Gene Expression Omnibus repository under accession code GSE1297 and GSE159699. A version of the Alzheimer's disease (AD) genomics data can be visualized at http://www.alzdata.org/. And protein expression in brain can be visualized at https://mouse.brain-map.org/ and http://www.proteinatlas.org/. All original data are available on request. All the other data supporting the findings of this study are available in the article and Inventory of Supporting Information files. Source data are provided with this article.
This article contains supporting information (46, 47, 62).
The authors declare that they have no conflicts of interest with the contents of this article. |
|
PMC9647553 | Qian Guo,Yaqi Qiu,Yiwen Liu,Yiqing He,Guoliang Zhang,Yan Du,Cuixia Yang,Feng Gao | Cell adhesion molecule CD44v10 promotes stem-like properties in triple-negative breast cancer cells via glucose transporter GLUT1-mediated glycolysis | 12-10-2022 | CD44v10,GLUT1,glycolysis,stem cell-like property,triple-negative breast cancer,ALDH, aldehyde dehydrogenase,CSC, cancer stem cell,CD44s, standard CD44 form,CD44v, variant CD44 isoforms,GLUT1, glucose transporter 1,ROS, reactive oxygen species,TNBC, triple-negative breast cancer | Cell adhesion molecule CD44v8-10 is associated with tumor ste0mness and malignancy; however, whether CD44v10 alone confers these properties is unknown. Here, we demonstrated that CD44v10 promotes stemness and chemoresistance of triple-negative breast cancers (TNBCs) individually. Next, we identified that genes differentially expressed in response to ectopic expression of CD44v10 are mostly related to glycolysis. Further, we showed that CD44v10 upregulates glucose transporter 1 to facilitate glycolysis by activating the MAPK/ERK and PI3K/AKT signaling pathways. This glycolytic reprogramming induced by CD44v10 contributes to the stem-like properties of TNBC cells and confers resistance to paclitaxel treatment. Notably, we determined that the knockdown of glucose transporter 1 could attenuate the enhanced effects of CD44v10 on glycolysis, stemness, and paclitaxel resistance. Collectively, our findings provide novel insights into the function of CD44v10 in TNBCs and suggest that targeting CD44v10 may contribute to future clinical therapy. | Cell adhesion molecule CD44v10 promotes stem-like properties in triple-negative breast cancer cells via glucose transporter GLUT1-mediated glycolysis
Cell adhesion molecule CD44v8-10 is associated with tumor ste0mness and malignancy; however, whether CD44v10 alone confers these properties is unknown. Here, we demonstrated that CD44v10 promotes stemness and chemoresistance of triple-negative breast cancers (TNBCs) individually. Next, we identified that genes differentially expressed in response to ectopic expression of CD44v10 are mostly related to glycolysis. Further, we showed that CD44v10 upregulates glucose transporter 1 to facilitate glycolysis by activating the MAPK/ERK and PI3K/AKT signaling pathways. This glycolytic reprogramming induced by CD44v10 contributes to the stem-like properties of TNBC cells and confers resistance to paclitaxel treatment. Notably, we determined that the knockdown of glucose transporter 1 could attenuate the enhanced effects of CD44v10 on glycolysis, stemness, and paclitaxel resistance. Collectively, our findings provide novel insights into the function of CD44v10 in TNBCs and suggest that targeting CD44v10 may contribute to future clinical therapy.
Triple-negative breast cancer (TNBC) is the most lethal subtype of breast cancers (BrCas), in which the higher frequency of cancer stem cells (CSCs) correlates with a dismal prognosis. Despite the current standard cure modalities, such as surgery, chemotherapy, and radiotherapy, most TNBC patients inevitably suffer from drug resistance and tumor recurrence. Therefore, a better understanding of the molecular basis for TNBC progression is urgently needed to discover new and more effective therapeutic targets for patients. CD44 is a major adhesion molecule and has been implicated in various biological behaviors, such as cell differentiation and cell motility, as well as tumor growth and metastasis (1, 2). The human CD44 gene consists of 19 exons and undergoes extensive alternative splicing that generates two families of isoforms, the standard CD44 form (CD44s) and the variant CD44 isoforms (CD44v). In contrast to the ubiquitous expression of CD44s, CD44v, which contains one or more variant exons, seems to be restricted to subpopulations endowed with stem cell potential and tumor development. Among CD44v, CD44v10-containing isoforms encompass a group of isoforms, such as CD44v8-10, CD44v3-10, and CD44v2-10, which have been identified as the most abundant variants (2, 3). It has also been reported that CD44v10-containing isoforms are involved in regulating metastasis, stemness, and chemoresistance of several cancers (4, 5, 6). Blockade of CD44v10 could downregulate the expressions of all isoforms containing exon-v10 and consequently delay tumor growth and metastasis (7, 8, 9). Therefore, targeting CD44v10-positive cancer cells arises as a promising cancer therapy option. However, the question of whether it is CD44v10 alone or CD44v10 cooperates with its related isoforms together that influences cancer behaviors remains unanswered. In this study, we first demonstrated that CD44v10 is preferentially expressed in TNBC patients and correlates with tumor progression. Next, we generated a breast cancer cell line transfected with CD44v10 cDNA and identified that glycolysis-related genes were significantly upregulated in response to CD44v10 overexpression. As reported before, abnormal glycolysis metabolism is usually adopted by CSCs, such as glucose uptake, glycolytic enzyme expression, and lactate production which are elevated in CSCs as compared with their differentiated offspring (10, 11, 12, 13). Moreover, certain CD44v isoforms, such as CD44v3, CD44v6, and CD44v8-10, could act as CSCs markers and play critical roles in promoting the properties of CSCs (5, 14, 15). These findings raise the possibility that CD44v10 may be involved in regulating the stem-like features. We next performed transcriptomic analysis and determined that CD44v10 enhances glycolysis through glucose transporter 1 (GLUT1) upregulation, which could maintain TNBC cells stem-like properties and confer paclitaxel (PTX) resistance. Notably, silencing of GLUT1 could markedly suppress the enhanced effects of CD44v10 on glycolysis and stem-like properties. Taken together, our findings suggested that CD44v10 could promote BrCas stemness individually without cooperating with other CD44 variants. The underneath mechanism of CD44v10-GLUT1 may point to CD44v10 as a specific therapeutic target to the clinic in TNBC.
To investigate the potential role of CD44v10 in BrCas, we first analyzed its expression in a cohort of 25 paired BrCas tissues. The results of immunohistochemical staining showed that CD44v10 levels were dramatically elevated in cancer tissues than adjacent noncancerous tissues (Fig. 1A). Next, to identify the correlation between CD44v10 and BrCas heterogeneity, we detected CD44v10 expressions among distinct molecular subtypes by using another cohort, which contains 30 cases of luminal BrCas, 13 cases of BrCas with HER2 overexpression, and 28 cases of TNBCs. Our results revealed that CD44v10 expression was significantly elevated in TNBCs compared with that in HER2-overexpression and luminal BrCas (Fig. 1B), implicating an important role of CD44v10 in the carcinogenesis of TNBCs. Similar results were obtained by utilizing several human BrCa cell lines showing that CD44v10 expression was higher in four TNBC cell lines (MDA-MB-468, MDA-MB-231, Hs578t, and BT-549) than human normal breast epithelial cells (MCF 10A) (Fig. 2A). Subsequently, we analyzed the correlation between CD44v10 and clinicopathological features in cohort 1 and cohort 2 BrCa patients. We did not find an obvious correlation of CD44v10 expression with age, tumor size, and lymph node status; however, tumors with high CD44v10 expression had higher histopathological grades and more advanced Tumor, Lymph Node, Metastasis stages (Table 1). Moreover, Kaplan–Meier curves were measured in an additional tissue microarray containing 135 human BrCas to identify the correlation between CD44v10 and survival time. The results revealed that BrCas patients with higher expression of CD44v10 showed shorter overall survival time (Fig. S1). Collectively, these data suggested that CD44v10 is preferentially expressed in TNBC patients and may play a vital role in the development of human BrCas.
Our previous study showed that expressions of all isoforms containing exon-v10 were significantly decreased in MDA-MB-231 cells after CD44v10 siRNA transfection (7). In the present study, similar results were obtained by using other TNBC cell lines (BT-549, Hs578t, and MDA-MB-468) (Fig. 2B). To determine the roles of CD44v10 itself in TNBC progression, we established a stable cell line by transfection of CD44v10 cDNA into BT-549 cells which contains little endogenous CD44v10 (Fig. 2C). The global transcriptome profiles in BT-549 cells transduced with vector or CD44v10 were examined by RNA sequencing. Differential expression analysis (fold change >1.5, p < 0.05) highlighted 898 and 1164 genes significantly upregulated and downregulated in BT-549 CD44v10 cells compared with vector samples, respectively (Fig. 2D). Functional annotation revealed that genes upregulated in CD44v10 overexpression cells were significantly enriched for pathways related to glycolysis, gluconeogenesis, and angiogenesis (Fig. 2E). To further extend this observation, we applied gene set enrichment analysis and confirmed a significant positive enrichment of glycolysis gene sets in BT-549 CD44v10 group compared with BT-549 vector cells (Fig. 2F). Taken together, our results suggested that CD44v10 may be involved in glycolysis.
To validate the RNA sequencing results, we next examined the changes in glycolytic metabolism altered by CD44v10 expression. In addition to BT-549 cells, we also knocked down CD44v10 in MDA-MB-231 cells, which contain a high level of endogenous CD44v10 (Fig. 3A). Given that glycolytic phenotype is characterized by increased glucose consumption and lactate secretion along with reduced reactive oxygen species (ROS) levels (16), we examined whether alteration of CD44v10 expression affects levels of glucose, lactate, and ROS. Our results indicated that ectopic expression of CD44v10 increases glucose consumption and lactate secretion as well as decreases intracellular ROS level, while the knockdown of CD44v10 reverses such effects (Fig. 3, B–D). These findings suggested that CD44v10 could promote the glycolytic processes of TNBC cells.
To gain insight into the mechanism by which CD44v10 potentiated TNBC glycolytic phenotypes, the gene expression patterns of key glycolytic enzymes in the glucose metabolic pathway were compared between BT-549 vector and BT-549 CD44v10 overexpression groups. As expected, most of the glycolysis-related genes were upregulated in CD44v10 overexpression cells (Fig. 4A). We further validated these gene expression levels by qPCR assay and selected three candidates (GLUT1, PGK1, and ENO2) that were robustly upregulated upon CD44v10 overexpression (Figs. 4B and S2). Next, the protein expressions of the three genes were detected, and the results showed that GLUT1, the key rate-limiting factor of cellular glucose uptake, was significantly increased in CD44v10 overexpression cells, while no obvious changes of PGK1 and ENO2 were observed upon CD44v10 ectopic expression (Figs. 4C and S2). These results suggested that GLUT1 may be principally responsible for the CD44v10-induced glycolysis. To substantiate the contribution of GLUT1 in CD44v10-mediated glycolysis, we next investigated whether the knockdown of GLUT1 affects the activated glycolytic processes caused by CD44v10 overexpression. The CD44v10 overexpression cells were transfected with control siRNA or GLUT1 siRNA (Fig. 4D). As shown in Fig. 4E, the silence of GLUT1 attenuated the increased levels of glucose consumption and lactate production induced by CD44v10 overexpression. A similar result was also obtained on ROS generation, which showed that the knockdown of GLUT1 rescued the reduced ROS level caused by CD44v10 overexpression (Fig. 4F). Further, we explored the mechanism by which CD44v10 induces GLUT1 expression. As we previously reported that CD44v10 could activate MAPK/ERK and PI3K/AKT signaling pathways (7) and others have indicated that such signaling could induce the GLUT1 expression (17, 18), we next asked whether MAPK/ERK and PI3K/AKT cascades are engaged in the CD44v10-induced upregulation of GLUT1. As expected, the inhibition of the signaling pathways could impair the CD44v10-triggered GLUT1 upregulation (Fig. 4G). Collectively, the data suggested that CD44v10 may upregulate GLUT1 via activating MAPK/ERK and PI3K/AKT signaling pathways in facilitating glycolysis.
Metabolic reprogramming in cancer cells has been reported to support the maintenance of CSCs properties (12). Additionally, it is now accepted that CD44v has been identified as one of the important markers of CSCs in many malignant tumors, which prompted us to further investigate whether CD44v10-activated glycolysis could contribute to the stemness properties of TNBC cells. After carrying out sphere formation assays, we found that CD44v10 overexpression remarkably increased the formation of spheroids compared to the control cells in BT-549 cells, whereas CD44v10 knockdown greatly hindered the sphere numbers (Fig. 5A). As the knockdown of CD44v10 could interfere the expressions of all CD44v10-containing isoforms, we next performed a rescue experiment by re-expressing CD44v10 in MDA-MB-231 shCD44v10 cells to further verify the role of individual CD44v10 in mediating TNBC stemness. The results showed that the re-expression of CD44v10 could restore the reduced sphere-forming ability of MDA-MB-231 caused by CD44v10 knockdown (Fig. S3). In addition, Oct4, Klf4, c-Myc, aldehyde dehydrogenase (ALDH)1, and Nanog are essential renewing and pluripotent regulators of stem cells and CSCs (19). We next examined the effects of CD44v10 on the expressions of these CSC-related factors and found that their expressions were upregulated in BT-549 CD44v10 overexpression cells while downregulated in the MDA-MB-231 shCD44v10 group (Fig. 5B). Furthermore, ALDH-positive populations are believed to contain breast CSC subsets (20). In line with these observations, the proportions of ALDH-positive cells increased from 13.00% to 23.63% in BT-549 CD44v10 overexpression compared with control cells, whereas it decreased from 12.12% to 6.75% in MDA-MB-231 siCD44v10 compared with sicontrol cells (Fig. 5C). To further corroborate the function of CD44v10 on stemness, we then analyzed the expression of CD44v10 in TNBC cell lines grown as either adherent monolayers or nonadherent spheres enriched in CSCs. The results showed that both mRNA and protein levels of CD44v10 were elevated in spherical cells compared to their adherent cells (Fig. 5, D and E). To confirm the contribution of GLUT1 in CD44v10-mediated stemness, we investigated whether the knockdown of GLUT1 could attenuate the enhanced stemness caused by CD44v10 overexpression. As shown in Fig. 5, F and G, GLUT1 knockdown abrogated the upregulation of tumorsphere and stemness gene expression induced by elevated expression of CD44v10 in BT-549 cells. Taken together, these results indicated that CD44v10 confers TNBC stem cell-like phenotype by upregulating GLUT1 expression.
CSCs are inherently responsible for tumor resistance to conventional chemotherapy. PTX is a predominantly used systemic treatment for TNBC patients. In light of the crucial role of CD44v10 in CSCs regulation, we sought to investigate whether the CD44v10 isoform is associated with chemosensitivity to PTX. Western blot results showed that PTX treatment increased the expression of CD44v10 along with CSC-related genes (Oct4, Klf4, ALDH, c-Myc, and Nanog) in a dose-dependent manner in both BT-549 and MDA-MB-231 cell lines (Fig. 6A). In addition, concerning chemical sensitivity, the cytotoxic effects of different concentrations of PTX on CD44v10-regulated cells and their respective control cells were detected after 48 h treatment of PTX by CCK-8 assays. The results indicated that the upregulation of CD44v10 in BT-549 cells resulted in resistance to PTX and its knockdown in MDA-MB-231 cells enhanced drug sensitivity to PTX-induced growth inhibition (Fig. 6B). To further verify the role of individual CD44v10 in regulating PTX sensitivity, we performed rescue experiments and found that the re-expression of CD44v10 could attenuate the increased PTX sensitivity caused by CD44v10 knockdown (Fig. S4). Similar results were also obtained in plate colony formation assay (Fig. 6C). These findings indicated that CD44v10 is associated with tumor resistance to chemotherapy. To confirm the contribution of glycolysis in CD44v10-mediated PTX sensitivity, we downregulated GLUT1 in BT-549 CD44v10 overexpression cells and found that the knockdown of GLUT1 could rescue the reduced PTX sensitivity caused by CD44v10 overexpression (Fig. 6D).
CD44v10-containing isoforms have been concerned for their overt activities in tumor growth, metastasis, and stemness. However, little is known about whether CD44v10 itself plays role in cancer progression compared to others such as CD44v6 and CD44v3. In this study, we demonstrated that CD44v10 independently promotes stem-like properties and decreases PTX sensitivity of TNBC cells at in vitro experiments. More importantly, we reported a novel mechanism of CD44v10 on BrCas stemness through glucose metabolism, which is mediated by GLUT1-induced glycolysis. Although high-level expression of CD44v10 has been observed in several cancers and correlates with a worse prognosis (21, 22, 23), there has limited information regarding the carcinogenic role of CD44v10 in BrCas. Herein, we evaluated the clinical significance of CD44v10 in breast cancer patients and found that CD44v10 was preferentially expressed in TNBC patients compared with HER2-positive or luminal BrCas. Notably, BrCa patients with high CD44v10 expression had higher histopathological grades and Tumor, Lymph Node, Metastasis stages, as well as shorter overall survival time, implying a potential role of CD44v10 in BrCas progression. Accumulating evidences have indicated that CD44v10-containing isoforms, such as CD44v8-10, CD44v4-10, and CD44v3,8 to 10, are related to cancers in diversity (4, 5, 24). However, the function of CD44v10 alone was largely unknown. Given the fact that the knockdown of exon-v10 could downregulate all expressions of CD44v10-containing isoforms (7), we performed transcriptomic analysis by establishing a CD44v10 overexpression cell line instead of CD44v10 knockdown to search for mechanisms on CD44v10 in regulating BrCas progression. We found that the differentially expressed genes induced by ectopic expression of CD44v10 were mostly associated with glycolytic processes. We then confirmed that CD44v10 upregulation could promote glycolytic phenotypes, including glucose consumption and lactate production along with ROS defense, whereas CD44v10 downregulation reverses such effects. These findings were consistent with previous studies showing that CD44 is involved in the regulation of the glycolytic pathway in cancer cells (25, 26). To further investigate the deeper mechanism of CD44v10-mediated glycolysis, we focused on glycolysis-related genes. After transcriptomic analysis of TNBC cells, we identified that GLUT1 was the most robustly upregulated gene in CD44v10 overexpression cells. As illustrated before, GLUT1 is the main transporter for cellular glucose uptake and has been reported to be significantly elevated in TNBCs (27). Our study firstly connected GLUT1 with CD44v10 for the findings that inhibiting GLUT1 could attenuate CD44v10-enhanced glycolysis in TNBC cells. However, it should be noted that GLUT1 downregulation did not completely disrupt the glycolytic phenotype, we hypothesized that CD44v10 may facilitate glycolysis through upregulation of GLUT1 expression along with other genes which need further confirmation. In supporting with our results, a previous study indicated that CD44 ablation could suppress GLUT1 expression (25) in which the underlying mechanism has not been clarified. Our lab recently reported that CD44v10 could activate MAPK/ERK and PI3K/AKT signaling pathways (7). Meanwhile, others demonstrated that the induction of GLUT1 was associated with the above signaling (17, 28). Therefore, we questioned whether CD44v10 could upregulate GLUT1 by triggering MAPK/ERK and PI3K/AKT pathways. Consistent with our hypothesis, the inhibition of the pathways significantly reduced CD44v10-induced GLUT1 protein expression, implying that CD44v10 may upregulate GLUT1 through the activation of MAPK/ERK and PI3K/AKT signaling pathways. It has been well accepted that metabolic reprogramming is one of the hallmarks of cancer (29). Recent studies suggested that CSCs may have higher glycolytic activity compared with the bulk of the general cancer cells (12, 13). As CD44v8-10 and CD44v4-10 could facilitate CSC activities in gastric cancer and intestinal cancer (5, 24), we wonder whether CD44v10 function a crucial role in the regulation of TNBC stemness properties independently. Our following experiments revealed that elevated CD44v10 was not only enriched in spheres of TNBC cells but was also essential to inducing a stem cell-like state. These findings suggested that CD44v10 alone may confer stemness properties and potentially act as a marker of CSCs in TNBC. Notably, we also demonstrated that the disruption of GLUT1 decreases glycolytic activity and thereby depresses the enhanced stemness induced by CD44v10 upregulation, providing a novel mechanism by which CD44 variants confer stem-like properties of cancer cells. In light of these observations, CD44v10 may individually promote stem cell-like properties of TNBC cells through GLUT1-mediated glycolysis. A growing body of literature has indicated that CD44v-expressing cancer cells especially CSCs show inherent chemoresistance (30, 31). Given the potential role of CD44v10 in CSCs regulation, we next asked whether CD44v10 could play a role in mediating PTX sensitivity. We accordingly performed drug sensitivity experiments in vitro and found that TNBC cells treated with PTX expressed higher levels of CD44v10 along with CSC-related genes in a dose-dependent manner. Additionally, the upregulation of CD44v10 alone could inhibit the sensitivity to PTX-induced cell death, while downregulation of CD44v10 reverses this effect, suggesting the involvement of CD44v10 in the regulation of PTX sensitivity. In concordance with our results, a previous study indicated that CD44v8-10 attenuates apoptotic responses to cisplatin in urothelial cancer (6). Furthermore, other CD44 variant exons, such as CD44v6 and CD44v9, have also been described to confer chemotherapeutic resistance to cancer cells through the regulation of redox balance (32, 33, 34). In view of this, the results of our present study raise the possibility that CD44v10 potentiates the ability of TNBC cells to defend themselves against chemotherapy-induced ROS. Strikingly, we further confirmed that the attenuation of PTX sensitivity caused by CD44v10 overexpression could be rescued by GLUT1 downregulation. To this end, we proposed a notion that targeting CD44v10 may disrupt the glucose metabolism of CSCs and thereby improve the sensitivity of TNBC cells to therapeutics, which warrants further investigation. In conclusion, our study unveiled the roles of CD44v10 alone in promoting stemness properties and regulating PTX sensitivity through GLUT1-mediated glycolysis, indicating that CD44v10 may be a therapeutic target to improve TNBC therapy.
Human BrCa cell lines (MCF10A, MDA-MB-468, Hs578t, BT-549, and MDA-MB-231) were purchased from the Cell Bank of the Type Culture Collection of the Chinese Academy of Sciences. All human cell lines have been authenticated using STR profiling within the last 3 years. MCF10A cells were cultured in MEGM (Lonza, Thermo Fisher Scientific, Inc), BT-549 cells were cultured in RPMI-1640 medium (Gibco, Thermo Fisher Scientific, Inc), and MDA-MB-468, Hs578t, and MDA-MB-231 cells were cultured in high-glucose Dulbecco's modified Eagle's medium (Gibco, Thermo Fisher Scientific, Inc). All the media were supplemented with 10% FBS (Bovogen Biologicals Pty Ltd), 100 U/ml penicillin, and 100 mg/ml streptomycin. Additional insulin with a final concentration of 0.01 mg/ml was added to the culture medium of BT-549 and Hs-578t. All cell lines were maintained at a temperature of 37 °C in humidified air with 5% CO2. In addition, all cells were grown to 80% confluency for the experiments. All experiments were performed with mycoplasma-free cells. The ERK inhibitor U0126 (S1102) and PI3K inhibitor LY294002 (S1105) were purchased from Selleck. The standard stock and trial solutions were prepared according to the manufacturer’s instructions.
Three independent tissue microarrays of human BrCas were purchased from Shanghai Outdo Biotech. To investigate the expression of CD44v10 in human BrCas, a tissue microarray (HBre-Duc060CS-03) including 25 paired available breast cancerous tissues and peritumoral tissues was examined by immunohistochemistry. To further evaluate the relevance of CD44v10 in different subtypes of BrCas, another cohort (HBreD080CS01) that consisted of 71 available breast primary tumor tissues (30 cases of luminal BrCas, 13 cases of BrCas with HER2 overexpression, and 28 cases of TNBCs) and six adjacent normal tissues was analyzed. Additionally, a tissue microarray (HBreD136Su02) containing 135 available BrCa patients was prepared to identify the correlation between CD44v10 and survival time. Immunohistochemical staining for CD44v10 was performed as reported previously (35). Briefly, these tissue sections were treated as follows: dewaxing, dehydration, antigen retrieval, and inhibition of endogenous peroxidase activity as well as nonspecific binding. Then, a mouse anti-human CD44v10 (1:200, Bio-Rad, MCA1733) antibody was added, and the slides were incubated at 4 °C overnight. After washing with PBS, the slides were incubated with a biotinylated secondary mouse antibody for 1 h, followed by streptavidin-ABC at room temperature. Then, the slides were developed with a 2,4-diaminobutyric acid Substrate Kit and counterstained with hematoxylin. The intensities of CD44v10 were quantitatively analyzed using IMAGE-PRO PLUS 6.0 software (Media Cybernetics). Mean density = IOD/area.
Total RNA was extracted from cultured cells using TRIzol reagent. After quantification using a NanoDrop 2000 spectrophotometer, purified total RNA (1 μg) was reverse transcribed with the PrimeScript RT Reagent Kit with gDNA Eraser. Real-time PCR assays were performed by using SYBR Green Mix according to the manufacturer’s instructions. All quantitative real-time PCR values of each gene were normalized against that of β-actin. The relative expression of genes was calculated by the 2-ΔΔCt method. The primer sequences for quantitative real-time PCR are listed in Table S1.
RIPA buffer (Beyotime) was used for protein extraction. After the total protein concentration was determined by a bicinchoninic acid protein assay kit (Thermo Fisher Scientific, Inc), 20 μg protein samples were separated by 8% SDS polyacrylamide gels and transferred onto PVDF membranes (Millipore). The membranes were blocked with 5% skimmed milk in 0.1% TBS-Tween-20 at room temperature for 1 h and incubated with following primary antibodies overnight at 4 °C: CD44v10 (1:1000, Sigma-Aldrich, AB2082), pan-CD44 (1:1000, Abcam, ab189524), GLUT1 [1:1000, Cell Signaling Technology, (CST), 73,015], Oct4 (1:1000, CST, 2750), c-Myc (1:1000, CST, 5605), Klf4 (1:1000, CST, 4038), ALDH1 (1:1000, CST, 12,035), Nanog (1:1000, CST, 4903), p-ERK (1:2000, CST, 4370), ERK (1:1000, CST, 4695), p-AKT (1:1000, CST, 4060), AKT (1:1000, CST, 4691), and β-actin (1:1000, CST, 3700). On the following day, horseradish peroxidase-conjugated secondary antibodies (1:5000, Lianke Biotech Co., Ltd) were added. Then, the protein bands were subsequently captured using the enhanced plus chemiluminescence assay (Pierce) and were measured on an ImageQuant LAS 4000 mini.
The siRNA constructs targeting human CD44v10 (NM_001202555) expression were designed and synthesized by RiboBio company (Guangzhou, China), and transfection was performed with riboFECT according to the manufacturer’s protocol. Sequences for siRNA knockdown, including CD44v10 siRNA (5′-CUACUUUACUGGAAGG UUA-3′), GLUT1 siRNA (5′-CUGUGGGCCUUUUCGUUAA-3′), and the scramble negative control siRNA sequence was 5′-UUCUCCGAACGUGUCACGU-3’. CD44v10 siRNA construct was inserted into the lentiviral vector hU6-MCS-CBh-gcGFP-IRES-puromycin (Genechem Company) for lentivirus production. Human CD44v10 was constructed by PCR-based amplification and cloned into the Ubi-MCS-3FLAG-SV40-Cherry-IRES-puromycin vector system (Genechem Company). Cells were infected with concentrated lentivirus according to the manufacturer’s instructions and treated with puromycin (2 μg/ml; Santa Cruz Biotechnology) for 3 weeks to generate stable CD44v10-knockdown or CD44v10-overexpression cells. The knockdown efficiencies and stable overexpression were verified by Western blotting assay.
Total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer’s protocol. The RNA libraries were then sequenced using an Illumina Novaseq 6000 instrument at the OE Biotech Co., Ltd. FPKM (Fragments Per kb Per Million Reads) of each gene was calculated using Cufflinks, and the read counts of each gene were obtained by HTSeq-count. Differential expression analysis was performed using the DESeq (2012) R package. We utilized a fold change > 1.5 with p-value < 0.05 as the threshold for significantly differential expression. Gene ontology enrichment and KEGG pathway enrichment analysis of differentially expressed genes were performed respectively using R based on the hypergeometric distribution. Gene set enrichment analysis (https://www.broadinstitute.org/gsea/index.jsp) was performed to analyze whether a set of genes show statistically significant differences between CD44v10 overexpression and vehicle control groups.
Glucose and lactate concentrations of the cultured medium were measured by using a Glucose Assay kit-WST and Lactate Assay kit-WST (Dojindo Laboratories) according to the manufacturer’s instruction. Intracellular levels of ROS were determined by using 2′,7′-dichlorofluorescein-diacetate (Beyotime Biotech) according to the manufacturer’s instructions.
Cells were seeded into ultralow attachment 6-well plates (Corning) at a density of 1 × 104 cells per well and then cultured in Dulbecco's modified Eagle's medium/F12 medium supplemented with B27 (1:50, Invitrogen), 20 ng/ml human recombinant epidermal growth factor (Peprotech), and 20 ng/ml human recombinant basic fibroblast growth factor (Peprotech) for 14 days. Cells were maintained in 5% CO2 at 37 °C, and the medium was replaced every 3 days. The total number of spheres greater than 50 μm in diameter was counted under an inverted microscope (Olympus Corporation).
The ALDEFLUOR kit (StemCell Technologies) was used to measure intracellular ALDH enzymatic activity. Briefly, one million cells were suspended in ALDEFLUOR assay buffer containing the ALDH substrate and incubated at 37 °C for 45 min. As a negative control, an aliquot of each cell sample was treated with diethylaminobenzaldehyde, a specific ALDH inhibitor. Stained cells were analyzed on the FACS CytoFLEX flow cytometer (Beckman-Coulter, Inc), and data analysis was performed using Cytexpert software 2.4 (Beckman-Coulter, Inc).
The sensitivity of cells to PTX was measured by CCK-8 assay (Dojindo Laboratories). Cells were seeded in 96-well plates at a density of 3000 cells per well and then treated with various concentrations (2.5 nM, 5 nM, 10 nM 20 nM, or 40 nM) of PTX (MedChemExpress) after the cells were attached. Then incubated for 48 h, replaced fresh culture medium, and added CCK-8 reagent to each well according to the manufacturer’s protocol, incubated at 37 °C for an additional 2 h. Absorbance was measured at 450 nm using a microplate absorbance reader (Bio-Rad Laboratories, Inc). Colony formation assay of BrCas cells was carried out by plating infected cells at a density of 1000 cells/well in a 6-well plate, and 2 nM of PTX was added. After 2 weeks of incubation, cells were washed three times with PBS, fixed with 4% paraformaldehyde for 30 min at room temperature, followed by staining with 0.1% crystal violet (BaSO Biotech Co., Ltd) for 30 min at room temperature. After rinsing three times, the stained colonies were imaged, and the number of colonies was counted by the naked eye.
Statistical analysis was performed using GraphPad Prism 8.0.2 (GraphPad Software, Inc). Data are presented as the group mean ± standard deviation (SD). Statistical analysis was performed using an unpaired Student’s t test for two-group comparison and a one-way analysis of variance (ANOVA) for multigroup comparisons. Differences in survival were calculated by the log-rank Mantel-Cox test. The p-value <0.05 was considered to be statistically significant (∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001). Generally, all experiments were carried out with triplicate independent replicates.
The data that support the findings of this study are available in the methods and/or supplementary material of this article.
This article contains supporting information.
The authors declare no conflict of interest with the contents of the article. |
|
PMC9647582 | Yu-Han Wen,Po-I Hsieh,Hsin-Cheng Chiu,Chil-Wei Chiang,Chun-Liang Lo,Yi-Ting Chiang | Precise delivery of doxorubicin and imiquimod through pH-responsive tumor microenvironment-active targeting micelles for chemo- and immunotherapy | 03-11-2022 | Chemotherapy,Immunotherapy,Tumor microenvironment,pH-sensitive,Micelles | Recently, combining immunotherapy and chemotherapy has become a promising strategy to treat cancer. However, this therapeutic strategy still has its limitations because of the adverse effects caused by the simultaneous administration of multiple therapeutic agents. Using nanoparticles is an effective approach to successfully combine these therapies because they can reduce side effects, increase circulation time, and ensure the delivery of cytotoxic agents to tumor tissues. In this study, dual pH-sensitive and tumor microenvironment (TME)-active targeting micelles comprising poly(propyl methacrylate-co-glucosamine/histidine/doxorubicin) (P(PAA-co-GLU/HIS/DOX) and methoxy-poly(ethylene glycol)-block-poly( l -lysine) were prepared to encapsulate an immunomodulator, imiquimod (IMQ). Because these micelles can expose glucose targeting ligands at the TME and pH-dependently release IMQ and DOX, micelles effectively inhibit the growth of 4T1 cells selectively and highly accumulate in 4T1 cells as the pH decreased to 6.5. Moreover, in RAW 264.7 cells, these micelles prevent cell death and induce M1 macrophage polarization. In 4T1 orthotopic tumor-bearing mice, micelles not only exhibited high tumor accumulation, effective tumor inhibition, and fewer adverse effects, but also dramatically increased the number of mature dendritic cells, activate cytotoxic T cells, and polarize M1-like macrophages in tumor tissues. Overall, these micelles exhibit precise pH responsiveness and ideal drug delivery capabilities for combined chemo- and immunotherapy; these results significantly contribute to the future development of nanomedicines in cancer therapy. | Precise delivery of doxorubicin and imiquimod through pH-responsive tumor microenvironment-active targeting micelles for chemo- and immunotherapy
Recently, combining immunotherapy and chemotherapy has become a promising strategy to treat cancer. However, this therapeutic strategy still has its limitations because of the adverse effects caused by the simultaneous administration of multiple therapeutic agents. Using nanoparticles is an effective approach to successfully combine these therapies because they can reduce side effects, increase circulation time, and ensure the delivery of cytotoxic agents to tumor tissues. In this study, dual pH-sensitive and tumor microenvironment (TME)-active targeting micelles comprising poly(propyl methacrylate-co-glucosamine/histidine/doxorubicin) (P(PAA-co-GLU/HIS/DOX) and methoxy-poly(ethylene glycol)-block-poly(l-lysine) were prepared to encapsulate an immunomodulator, imiquimod (IMQ). Because these micelles can expose glucose targeting ligands at the TME and pH-dependently release IMQ and DOX, micelles effectively inhibit the growth of 4T1 cells selectively and highly accumulate in 4T1 cells as the pH decreased to 6.5. Moreover, in RAW 264.7 cells, these micelles prevent cell death and induce M1 macrophage polarization. In 4T1 orthotopic tumor-bearing mice, micelles not only exhibited high tumor accumulation, effective tumor inhibition, and fewer adverse effects, but also dramatically increased the number of mature dendritic cells, activate cytotoxic T cells, and polarize M1-like macrophages in tumor tissues. Overall, these micelles exhibit precise pH responsiveness and ideal drug delivery capabilities for combined chemo- and immunotherapy; these results significantly contribute to the future development of nanomedicines in cancer therapy.
Malignant neoplasms are deadly diseases that take millions of lives around the world. Conventional treatments including surgery, chemotherapy, and radiation are commonly used for cancer therapy. However, these therapeutic strategies often cause adverse effects, making it necessary to develop an efficient way to treat cancer. In recent years, several immunotherapeutic products have been approved for use in cancer [1]. However, developing a new drug is riskier and has a higher cost than developing a drug delivery system or a new combined therapeutic strategy. Compared to an entirely new structure drug application, through 505(b) (2) new drug applications (NDAs) allow the use of published experimental results to avoid repeating studies and offer accelerated nonclinical testing programs [2]. Targeting the tumor microenvironment (TME) is a prudent strategy to not only minimize adverse effects during the treatment, but also improve therapeutic efficacy [[3], [4], [5]]. Numerous studies focused on drug delivery to the TME have reported dramatic tumor growth inhibition by a combination of immunotherapy with chemotherapy [6], radiation [7], or photothermal therapy [8]. The TME comprises cancer cells and various immunosuppressive cells, such as myeloid-derived suppressor cells (MDSCs), tumor-associated macrophages (TAMs), and T regulatory cells (Tregs) [9]. Cancer cells have a survival feature to evade immune recognition and exclusion [10]. Tumor immunosuppressive cells also contribute to immunosuppression and immune escape [9]. Activation of dendritic cells (DCs) can be useful in cancer treatments because they are crucial in linking the innate and adaptive immune responses [11]. Imiquimod (IMQ), a Toll-like receptor 7 (TLR-7) agonist, was used to treat actinic keratosis (AK) and superficial basal cell carcinoma (BCC) [12]. Once IMQ interacts with TLR-7 in the intracellular endosomes, DCs can begin eliciting cytokines and chemokines including IFNα, TNF-α, IL-2, IL-6, IL12, G-CSF, and GM-CSF. DCs could then be activated to attract mature cytotoxic T-lymphocytes and other immune cells into the tumor tissues, especially through transcription factors for NF-kB [13,14]. Moreover, IMQ has been shown to decrease Tregs recruitment and potentially switch TAMs into M1-like macrophages in the tumor tissue [15,16], resulting in tumor growth inhibition [17]. Although IMQ is an immunostimulant for immunotherapy, its clinical applications are highly limited because of its hydrophobicity and low tumor-targeting ability [6,16]. Nanoparticles are considered a potential therapeutic approach to improving drug dissolution, reducing adverse effects, and prolonging the circulation time of active pharmaceutical ingredients, thereby enhancing bioavailability and improving the therapeutic index [[18], [19], [20]]. Furthermore, nanoparticles are designed using materials well-suited for responding to the hallmarks of malignant neoplasms; therefore, they can specifically target and treat cancer cells. Cancer cells exhibit high glucose metabolism and consumption even in aerobic conditions through aerobic glycolysis, which is known as the “Warburg effect” [21,22]. This process also produces excess lactate, inducing a more acidic TME compared to normal tissues [23,24]. Based on these features of cancer cells, glucose-conjugating nanoparticles have been shown to have increased uptake by cancer cells, thereby exhibiting higher anti-tumor efficacy [25,26]. Moreover, controlling the release of the cargo from nanoparticles into tumor tissues or cancer cells has been achieved by developing pH-sensitive structures, such as histidine, acrylic acid [27], cis-aconityl amide [28], benzoic imine [29], hydrazone [30], and acetal/ketal [31] to specifically release encapsulated drugs into the TME or into intracellular endosomes/secondary lysosomes through structural protonation or hydrolysis. Based on this combined therapeutic strategy and drug interaction characteristics, a precise pH-responsive and TME-targeting micelle was designed to expose glucose ligands for enhanced uptake by cancer cells and the subsequent release of IMQ to induce TAM polarization and DC maturation, thereby achieving combined chemo- and immuno-therapy. In this study, two pH-responsive copolymers, poly(methyl methacrylate-co-glucosamine/histidine/doxorubicin) (P(MAA-co-GLU/HIS/DOX), abbreviated as M-HGD) and poly(propyl methacrylate-co-glucosamine/histidine/doxorubicin) (P(PAA-co-GLU/HIS/DOX), abbreviated as P-HGD) (Fig. 1A), with different pKa values of approximately 5.0 and 6.5 for MAA and PAA, respectively, were synthesized by free radical polymerization. The carboxylic groups of MAA and PAA were important pH-sensitive functional group. The NHS group of MAA-NHS was used to conjugate histidine, glucosamine and DOX. The chemotherapeutic drug, DOX, was conjugated by a pH-sensitive hydrazone bond on these copolymers. IMQ was then encapsulated with the M-HGD and P-HGD copolymers to form glucose-conjugating nanoparticles, which were then coated with a second polymer, methoxy-poly(ethylene glycol)-block-poly(l-lysine) (mPEG-b-PLys), synthesized by open-ring polymerization, through electrostatic interactions to form the micelles, hereafter denoted as ML-HGD and PL-HGD. In the TME, the carboxyl groups of PAA and imidazole groups of histidine on the PL-HGD could be protonated because of the pH changing from 7.4 to 6.5. Subsequently, mPEG-b-PLys could be desorbed from micelles to expose glucose ligands and partially release IMQ (Fig. 1B). The physically encapsulated IMQ was design for activating immune cells in TME, otherwise chemically conjugated DOX was plan to delivering to cancer cells. The exposed glucose ligands could significantly improve P-HGD nanoparticle delivery to cancer cells through endocytosis. The acidic surroundings of the endosomes induce the release of DOX from the copolymers through the hydrolysis of pH-sensitive hydrazone bonds for effective chemotherapy. Furthermore, P-HGD nanoparticles can also be internalized into TAMs and release the remaining IMQ in the phagosomes/phagolysosomes, thereby inducing the secretion of cytokines and the polarization of TAMs into M1-like macrophages for macrophage-mediated immunotherapy. Meanwhile, IMQ could be rapidly released into the TME because of the pH-sensitive characteristics of PL-HGD. A portion of the released IMQ could be absorbed by DCs in the TME. Afterward, the IMQ released in the TME could further mature the DCs and the cytotoxic T cells for cancer immunotherapy (Fig. 1C). Conversely, the PM-HGD performance lacked therapeutic efficacy on combinational therapy because the responsive site was in the late endosomes and secondary lysosomes instead of TME.
Amino functionalized methoxyl polyethylene glycol, amino PEG (mPEG5000-amine) was purchased from Nanocs (Boston, Massachusetts, USA). MAA, 3-mercaptopropionic acid (3-MPA), and sodium bicarbonate (NaHCO3) were obtained from Acros organics (Geel, Belgium). The 2,2′-Azobisisobutyronitrile (AIBN) was purchased from UniRegion Bio-Tech Inc (Taiwan).; Glucosamine hydrochloride (GLU), histidine (HIS), N-hydroxysuccinimide (NHS), N,N′-dicyclohexyl-carbodiimide (DCC), tert-butyl carbazate, hydrobromic acid solution (33 wt % in acetic acid), and trifluoroacetic acid (TFA) were purchased from Sigma-Aldrich (St. Louis, Missouri, USA). The DOX hydrochloride was purchased from LC Laboratories (Woburn, Massachusetts, USA) and IMQ was purchased from TCI (Tokyo, Japan). 4-Dimethylaminopyridine (DMAP) was purchased from Alfa Aesar (Ward Hill, Massachusetts, USA); Acetonitrile (ACN), dichloromethane (DCM), dimethylformamide (DMF), dimethyl sulfoxide (DMSO), methanol (MeOH), ethanol, diethyl ether, and isopropyl alcohol (IPA) were purchased from ECHO (Miaoli, Taiwan). N6-Carbobenzoxy-l-lysine N-carboxyanhydride (NCA-lysine) was purchased from Carbosynth (Compton, Berkshire, UK.). PAA was purchased from Polymer Source (Dorval, Quebec, Canada). AIBN was recrystallized from MeOH before use. DCM, DMF, and DMSO were dried and distilled with calcium hydride before use.
MAA (0.8 mL, 9.39 mmol), NHS (1.6230 g, 14.09 mmol), and DMAP (0.3445 g, 2.82 mmol) were dissolved in DCM (20 mL). A solution of DCC (3.8749 g, 18.78 mmol) in DCM (40 mL) was then added dropwise; the reaction was conducted at room temperature with stirring for 24 h. After the reaction was complete, acetic acid (0.4 mL, 6.99 mmol) was added to the mixture and incubated overnight at −20 °C. Reaction mixture was progressively filtered to remove 1,3-dicyclohexyl urea (DCU), extracted twice with saturated sodium bicarbonate solution and three times with deionized water. It was then evaporated in a rotary evaporator to obtain a white powder. The final product was recrystallized with IPA three times. 1H NMR (400 MHz, DMSO-d6, δ): 6.35 (s, 1H), 6.05 (s, 1H), 2.85 (s, 4H), 2.0 (s, 3H). FT-IR (KBr pellet): 2954 (O–H), 1761–1737 (C PBM data was replaced with SVG by xgml2pxml: <glyph-data id="pc-E00C" format="PBM" resolution="300" x-size="12" y-size="9"> 000000000000 000000000000 000000000000 111111111111 000000000000 111111111111 000000000000 000000000000 000000000000 </glyph-data> <svg xmlns="http://www.w3.org/2000/svg" version="1.0" width="20.666667pt" height="16.000000pt" viewBox="0 0 20.666667 16.000000" preserveAspectRatio="xMidYMid meet"><metadata> Created by potrace 1.16, written by Peter Selinger 2001-2019 </metadata><g transform="translate(1.000000,15.000000) scale(0.019444,-0.019444)" fill="currentColor" stroke="none"><path d="M0 440 l0 -40 480 0 480 0 0 40 0 40 -480 0 -480 0 0 -40z M0 280 l0 -40 480 0 480 0 0 40 0 40 -480 0 -480 0 0 -40z"/></g></svg> O), 1631 (CC), 1209–1087 (C–O).
P(PAA-co-NHS) and P(MAA-co-NHS) copolymers were synthesized by free radical polymerization [32]. Briefly, AIBN (7.6 mg, 0.046 mmol), PAA/MAA (190.2 mg, 1.667 mmol/54.5 mg, 0.621 mmol), MAA-NHS (169.6 mg, 0.896 mmol), and 3-MPA (24.6 mg, 0.232 mmol) were dissolved in a MeOH/DMSO co-solvent. The reactions were incubated for 24 h at 50 and 70 °C and then precipitated using cold diethyl ether to obtain P(PAA-co-NHS) and P(MAA-co-NHS), respectively. 1H NMR spectrum of P(PAA-co-NHS) (400 MHz, DMSO-d6, δ): 2.7–3.0 (br, 4H), 1.0–1.8 (br, 9H), 0.7–0.9 (br, 3H). FT-IR spectrum of P(PAA-co-NHS) (KBr pellet): 3446 (O–H), 2993–2962 (C–H), 1735–1676 (CO), 1203–1064(C–O). 1H NMR spectrum of P(MAA-co-NHS) (400 MHz, DMSO-d6, δ): 2.7–3.0 (br, 4H), 1.6–2.1 (br, 2H), 0.7–1.5 (br, 3H). FT-IR spectrum of P(MAA-co-NHS) (KBr pellet): 3410 (O–H), 2997−2947 (C–H), 1726−1718 (CO), 1215−1020 (C–O).
Either the P(PAA-co-NHS) or the P(MAA-co-NHS) copolymers were mixed with glucosamine hydrochloride in DMSO containing 10% TEA at 60 °C and incubated for 24 h. Then, histidine molecules were incubated with this mixture for 5 days for the reaction. After the reaction, excess amounts of tert-Butyl carbazate were added and incubated at 37 °C for 24 h. The mixture was then progressively precipitated using cold diethyl ether, reacted with TFA to remove the butyloxycarbonyl (boc)-protecting group for 2 h, and then conjugated with DOX at 37 °C for 24 h. The dark-red mixture was then placed in a dialysis bag (MWCO 1 K) against DMSO for 3 days and purified using a Sephadex LH-20 packing column to obtain P-HGD and M-HGD copolymers. The molecular weight and polydispersity index (PDI) were determined using gel permeation chromatography (GPC) system (LC-20AT, Shimadzu Co., Japan) with a Phenomenex Shodex OHpak ® SB-804 HQ column (10 μm, 300 mm × 8.0 mm), and an RI detector (RID-10 A, Shimadzu Co., Japan) at 40 °C. The mobile phase was DMF containing 50 mM lithium bromide with a flow rate of 1 mL/min 1H NMR of P-HGD (400 MHz, DMSO-d6, δ): 7.3–8.7 (br, 3H), 6.5–7.3 (br, 1H), 5.1–5.8 (br, 1H), 1.5–2.3 (br, 2H), 0.9–1.5 (br, 7H), 0.2–0.9 (m, 3H). FT-IR of P-HGD (KBr pellet): 3443 (O–H), 2976−2877 (C–H), 1720−1666 (CO), 1192−1016 (C–O). 1H NMR of M-HGD (400 MHz, DMSO-d6, δ): 7.1–8.8 (br, 3H), 6.6–7.1 (br, 1H), 5.2–5.6 (br, 1H), 1.5–2.2 (br, 2H), 0.3–1.5 (br, 3H). FT-IR of M-HGD (KBr pellet): 3441 (O–H), 2999−2949 (C–H), 1716−1676 (CO), 1188−1018 (C–O).
mPEG5000-amine was used as a macroinitiator for the ring-opening polymerization. Fixed amounts of mPEG5000-amine and NCA-lysine were dissolved DMF (50 mL) under nitrogen. The reaction was conducted at 35 °C for 2 days. The mixture was then progressively precipitated using ethanol, reacted with 33% HBr in acetic acid at room temperature to remove carboxybenzyl protecting group, and precipitated using cold diethyl ether to obtain mPEG-b-PLys. 1H NMR (400 MHz, D2O, δ): 4.1–4.2 (br, 26H), 3.4–3.7 (br, 455H), 3.2–3.3 (br, 3H), 2.8–3.0 (br, 27H), 1.1–1.8 (br, 162H). FT-IR (KBr pellet): 3771 (N–H), 2885 (C–H), 1656−1627 (CO), 1118 (C–O).
Fifteen mg of the M-HGD or the P-HGD copolymers were dissolved in 1.5 mL of DMSO and mixed with 2 mg of IMQ in 4 mL of MeOH. Core nanoparticles were then prepared using a solvent exchange process with a dialysis bag (MWCO 6–8 K) against deionized water. The water was changed every 12 h. After 3 days, the core nanoparticle solution was passed through a Sephadex G-50 packing column using phosphate buffer solution (PBS) with a pH of 7.4 as the elution solvent to remove free IMQ. The core nanoparticle solution was then mixed with 15 mg of mPEG-b-PLys and stirred for 24 h at room temperature to form micelles. Finally, the solution was progressively filtered through a 0.8-μm filter to remove precipitates and then centrifuged using an Amicon ultra centrifugal filter (30 K) to remove free mPEG-b-PLys. Particle size and polydispersity index (PDI) were measured for the micelles using dynamic light scattering (DLS, Zetasizer Nano ZS90, Malvern Panalytical, UK). The maximum absorption and emission spectrum was determined using UV/Vis spectrophotometer (Ultrospec 9000pc, Biochrom, United Kingdom) and plate reader (Infinite M200 Pro, TECAN, Switzerland), respectively. The morphologies of the micelles were observed by transmission electron microscopy (TEM, JEM-2000EXII, JEOL, Japan) at an accelerating voltage of 120 kV. In order to determine the stability of the micelles, sterile micelles were stored at 4 °C. At different time intervals, 10 μL of concentrated micelle solution was diluted with 1 mL of PBS; the size change was monitored for 2 months using DLS. In order to determine the pH-responsiveness of the micelles, 10 μL of the concentrated micelle solution was diluted with 1 mL of PBS at different pH levels (pH of 7.4, 6.5, and 5.0) and incubated at 37 °C. The hydrodynamic diameters of the micelles were measured at different time points using DLS. The accurate drug loading of IMQ was measured using a HPLC (Shimadzu Co., Japan) with an Inspire™ C18 column (5 μm, 250 mm × 4.6 mm), and an UV–Vis detector (SPD-10 A, Shimadzu Co., Japan) at 40 °C. The mobile phase comprised 70% sodium acetate (pH = 4) and 30% ACN with a flow rate of 1 mL/min. The quantitative analysis of the IMQ was performed at 244 nm. In contrast, the drug loading of DOX was measured by UV/Vis spectrophotometer at 480 nm. Drug loading measurement was calculated according to the formula as following:
In order to understand the IMQ- and DOX-releasing behaviors of the micelles, 1 mL of the micelle solution was transferred into a dialysis bag (MWCO 6–8 K) and immersed in 3 mL of PBS at different pH levels and incubated at 37 °C. At different time intervals, 3 mL of sample was collected, filtered through a 0.45-μm PVDF filter. The analyzed method was modified from HPLC procedure as mentioned before. HPLC equipped with UV–Vis detector (SPD 10 A, Shimadzu Co., Japan)/fluorescence detector (RF-10 AXL, Shimadzu Co., Japan). DOX was measured by detecting the fluorescence at an excitation and emission wavelength of 480 and 570 nm.
4T1 (8 × 104 cells/mL), RAW 264.7 (2 × 105 cells/mL), and L929 cells (1 × 105 cells/mL) were seeded on individual 96-well plates (each well contained 0.1 mL of cell suspension). After 12 h of incubation, various concentrations of free DOX, IMQ, DOX/IMQ, and micelles were added to the wells for 24 h. After removing drug/micelle-containing medium, the MTT assay was performed to determine the viability of each cell sample using an ELISA reader (Infinite M200 Pro, TECAN, Switzerland).
4T1 cells (5 × 105 cells/each well) were seeded on 6-well plates. After 12 h of incubation, cells were pretreated with 100 mg/mL of glucose for 1 h. Subsequently, either free DOX or micelles (at a DOX concentration of 1 μg/mL) along with 100 mg/mL of glucose were added to each well. After 3 h of co-incubation, the cells were harvested and washed with cold PBS. The intracellular fluorescence intensity for DOX was detected using flow cytometry (FACSCalibur, BD Biosciences, USA). For intracellular fluorescence observation of free DOX and micelles, 4T1 cells were seed in 18 × 18 mm cover glass. After 12 h, the medium was changed with various pH PBS containing 10% FBS and various formulated DOX. After 2 h, 1 μL of 50 μM LysoTracker Deep Red (Thermo Fisher; catalog: L12492) was added and treated for further 1 h. The cells were washed twice with cold PBS, fixed with 10% formalin and then sealed with DAPI-contained mounting gel. The intracellular distribution of different formulated DOX was observed by confocal fluorescence microscopy (LSM880, Zeiss, Germany).
RAW 264.7 cells were seeded at a density of 106 cells/well into 6-well plates. After 12 h, the culture medium was replaced with 2 mL of culture medium containing either free IMQ or micelles and incubated for another 6 h. Then, the cell culture supernatant was collected for TNF-α (eBioscience, catalog: 88-7324-88) and IL-6 (eBioscience, catalog: 88-7064-88) analysis using ELISA.
The method was modified from that used in a previous study [33]. In brief, fresh whole blood was collected from the mice in a heparin-containing tube. To remove the serum, the whole blood sample was centrifuged at 200 g for 5 min. The precipitated red blood cells (RBCs) were washed with sterile PBS five times and diluted with PBS after the last wash. A total of 0.2 mL of diluted RBC was mixed separately with either 0.8 mL of sterile PBS (negative control; NC), purified water (positive control; PC), free DOX, and micelles. Afterward, the mixture was incubated for 3 h. The supernatant was obtained by centrifugation at 2000 RPM for 10 min. The absorbance of hemoglobin in the supernatant was measured at 570 nm, and the reference was set at 620 nm. The hemolysis percentage was calculated according to the following equation:
Female BALB/c mice (4–6 weeks old) were purchased from the National Laboratory Animal Center (Taipei, Taiwan). All the animals were managed and experiments were performed in accordance with the Guidance on the Usage and Care of Laboratory Animals, approved by the institutional animal care and use committee (IACUC) of National Yang-Ming University. The 4T1 cells (1 × 106 cells in 100 μL of PBS) were translocated into the center of the right mammary fat pad of each mouse. Tumor volume was calculated using the following formula: tumor volume (mm3) = longest side × shortest side2/2.4T1 tumor-bearing BALB/c mice with tumor volumes of approximately 100–250 mm3 were administrated with Cy5.5-labeled micelles via tail vein injection. The organs and tumors were harvested and analyzed using in vivo imaging systems (PhotonIMAGER Optima, Biospace Lab, France) 24 h later. 4T1 tumor-bearing BALB/c mice with tumor volumes of approximately 100–250 mm3 were intravenously injected with free DOX, ML-HGD, and PL-HGD at a dose equivalent to 8 mg/kg of DOX. After 24 h, the mice were sacrificed and the tumor, liver, kidney, spleen, and lung were collected and embedded in Tissue-Tek® optimum cutting temperature (O.C.T.) solution with a liquid nitrogen bath. These tissues were cut to have a 10 μm thickness using a Cryostat Microtome (CM3050S, Leica, Germany). Next, the sliced tissue samples were sealed with mounting medium containing DAPI; the fluorescence intensities of DOX and DAPI were observed by confocal fluorescence microscopy (LSM880, Zeiss, Germany). For fluorescence image observation, the excitation/emission wavelength for DOX and DAPI was 480/595 nm and 405/450 nm, respectively.
The 4T1 cells (1 × 106 cells in 100 μL of PBS) were translocated into the center of the right mammary fat pad of each mouse. After 1 week of tumor implantation, mice with an average tumor volume of 50–200 mm3 were randomly assigned to 6 groups (each group contained 6 mice): PBS, DOX, IMQ, DOX + IMQ, PL-HGD, and 2-fold PL-HGD. The tumor-bearing mice were intravenously injected with 8 mg/kg of DOX and/or 25 μg/kg of IMQ (or PL-HGD at equivalent DOX and IMQ doses) four times, once every 3 days. Tumor volume and body weight were measured every 2 days. Mice were sacrificed on the 12th day from the first intravenous injection and whole blood samples were collected. Red blood cells (RBCs) and white blood cells (WBCs) were analyzed using a hematology analyzer (XT-1800iv, Sysmex, Japan). The remaining whole blood samples were centrifuged at 2000×g for 10 min to obtain plasma components for the evaluation of hepatic and renal function indexes such as glutamic pyruvic transaminase (GPT) and blood urea nitrogen (BUN) using an automated clinical chemistry analyzer (Fuji Dri-Chem 4000i, Fujifilm, Japan).
In order to evaluate the activation of DCs and polarization of macrophages, the right inguinal lymph nodes and tumors were isolated and dissociated into single cells by incubating with Accumax for 1 h. The cell suspension was passed through a 40-μm cell strainer to remove tissue masses and stained with PE Anti-Mouse CD86 (eBioscience, catalog: 17-0862-82), FITC Anti-Mouse CD80 (eBioscience, catalog: 11-0801-82), FITC Anti-Mouse CD11c (eBioscience, catalog: 11-0114-82), Alexa Fluor® 647 Anti-Mouse CD206 (Bio-Rad, catalog: MCA2235A647), FITC Anti-Mouse CD86 (Invitrogen, catalog: 11-0862-85), PE Anti-Mouse F4/80 (Elabscience, catalog: E-AB-F0995D), APC Anti-Mouse CD3 (BioLegend, catalog: 100,236) and FITC Anti-Mouse CD8 (Invitrogen, catalog: 11-0862-85). After 30 min, the cell suspension was centrifuged at 2000 rpm to collect the cells and then washed twice with cold PBS. The fluorescent marker was detected using a flow cytometer (Coulter CytoFLEX, Beckman coulter, USA). In order to observe CD3+ and CD8+ T cells in tumor tissues, the tumors were sequentially fixed with 3.7% paraformaldehyde, embed in parafilm, cut into 5-μm thick slices, and incubated with the primary antibody against anti-mouse CD3 (eBioscience, catalog: 14-0032-82), CD8 (eBioscience, catalog: 14-0808-82), and TNF-α (GeneTex catalog: GTX110520). Goat anti-mouse IgG and HRP-linked antibodies were the secondary antibodies used to bind the primary antibodies. Immunohistochemistry (IHC) images were observed using phase contrast microscopy (Invitrogen™ EVOS™ XL Imaging System, Thermo Fisher, USA). In order to evaluate the expression of iNOS, the tumor tissues were embedded in O.C.T. solution and frozen at −20 °C. These tissues were cut into 10-μm thick slices using a Cryostat Microtome (CM3050S, Leica, Germany). Next, the sliced tissue sample was blocked with PBS buffer containing 1% BSA and 0.1% Tween 20. After 30 min, the tissue slice was stained with diluted iNOS antibody (Miltenyi Biotec, catalog: 130-116-357) at 4 °C overnight. The sample was sealed with mounting medium containing DAPI and observed by confocal fluorescence microscopy (LSM880, Zeiss, Germany).
All experimental data were presented as the mean ± standard deviation (SD) from at least three independent experiments. The p-values were calculated by one-way ANOVA. A p-value lower than 0.05 indicated a statistically significant difference.
The number of carbons on acrylic acid monomer has been reported to affect the pKa values of acrylate polymers [34]. The pKa values of PAA and MAA have been reported to be similar to the extracellular and intracellular pH values in tumors [35], respectively; therefore, drug delivery systems comprising PAA and MAA can control the release of the drug both outside and inside the cancer cells. This enables the evaluation of the difference between the therapeutic efficacy of drug delivery systems on the TME and that on intracellular endosomes. Procedures for the synthesis of P-HGD and M-HGD copolymers have been illustrated in Fig. S1. The synthesized materials were characterized using 1H NMR and FT-IR to identify their structure and purity (Fig. S2–S6 and Table S1–S2). The average molecular weights and PDI of the copolymers were measured by GPC (Fig. S7) and summarized in Tables S1 and S2. The experimental results indicate that the molecular weight and polydispersity index (PDI) of the M-HGD and P-HGD copolymers were similar (Fig. S7B). From the 1H NMR analysis, the DOX contents (molar percentages) in the P-HGD and M-HGD copolymers were 23 and 19%, respectively, which were similar to those confirmed by UV–Vis spectrum (20.7% and 19.1%). In the result of titration, M-HGD copolymers exhibited two pKa values of approximately 4.5 and 6.8 contributed by the carboxylic acids of the methylacrylic acid and imidazole groups of histidine, respectively. In contrast, P-HGD copolymers showed a pKa value of 6.8 from histidine and propylacrylic acid (Fig. S8). The mPEG-b-PLys copolymers were synthesized by ring-opening polymerization, after which the carbobenzoxy protecting groups were removed to expose their positive charges. The structure of the copolymer was confirmed by analyzing the 1H NMR and FT-IR spectra (Fig. S9). The mPEG-b-PLys copolymers contained 27 Lys residues, as the repeating units. Copolymers become hydrophobic after conjugating with DOX; therefore, a solvent exchange process involving dialysis was used to prepare the nanoparticles and encapsulate a hydrophobic drug, IMQ. Core nanoparticles exhibited negative charges of approximately −30 mV because the pKa values of PAA and MAA were below the pH level of 7.4 (Table S3). The average particle size and PDI of the core nanoparticles (M-HGD and P-HGD) were larger than 130 nm and 0.2, respectively. The accumulation of nanoparticles with high negative charges in the liver has been shown to be significantly higher than those of slightly negative and neutral nanoparticles [36]; therefore, mPEG-b-PLys copolymers were used to shield the surface charges through electrostatic interactions between PLys and the MAA or PAA to form micelles. The DLS analysis results and drug loading summarized in Table 1 show that the average particle size and PDI of both micelles were decreased to approximately 60 nm and 0.15, respectively, because the electrostatic interactions compressed the core structures. The zeta-potentials for ML-HGD and PL-HGD also changed to 1.0 and 3.5 mv, respectively, indicating that the mPEG molecules were outside the micelles. Both drug loading of DOX and IMQ were similar in ML-HGD and PL-HGD micelles. The morphologies of micelles were observed by TEM, which revealed that both micelles had spherical structures and their sizes were similar to those measured by DLS (Fig. 2A and B; Fig. S10 and S11). The UV–Vis analysis results indicate that the absorption peak of DOX after conjugation on polymers shifted from 481 to 507 nm (Fig. S12A), indicating the formation of a hydrazone bond [37]. In addition, the maximum absorbance and fluorescence intensity of micelles dramatically increased when micelles were suspended in DMSO (Fig. S12B). The quenching effect was observed because of the spatial proximity of the DOX molecules. The long-term stability for ML-HGD and PL-HGD were monitored for 2 months at 4 °C. For PL-HGD, the average particle size and PDI were maintained at around 60 nm and 0.2, respectively. Although the size of ML-HGD did not change, the PDI slightly increased and showed a high deviation after 50 days (Fig. 2C), suggesting that it was less stable than PL-HGD in PBS. ML-HGD and PL-HGD contained pH-responsive materials such as MAA, PAA, and histidine molecules that allowed them to change their structures at different pH levels. The pH-triggered size changes were evaluated using various pH buffers to mimic blood circulation (pH 7.4), TME (pH 6.5 and 6.8), and intracellular endosomes/secondary lysosomes (pH 5.0). Both ML-HGD and PL-HGD exhibited stability at a pH of 7.4 in PBS (Fig. 2D–G). However, the particle size or PDI were significantly different at pH levels of 6.5, 6.0, and 5.0 for PL-HGD at 12 h (Fig. 2E,G). As time increased, the particle sizes and PDI notably increased at both pH 6.0 and pH 5.0 compared to those at pH 7.4. For ML-HGD, particle size and PDI were only varied at a pH of 5.0 (Fig. 2D,F). Form the results of TEM, ML-HGD exhibited integral structures when the pH changed from 7.4 to 6.0. However, an increased size of ML-HGD was observed at pH 5.0 due to structural deformation. In contrast to ML-HGD, PL-HGD exhibited a swollen and increased size in a mildly acidic environment, which was consistent with the results obtained with DLS measurement (Fig. S13). The zeta potential of PL-HGD was largely increased after 1 h at a pH of 6.0 in PBS, whereas ML-HGD retained their zeta-potentials even after 6 h at a pH of 6.0 (Fig. 2H). These experimental results indicate that PL-HGD could be protonated and desorbed from the mPEG-b-PLys at the TME (pH 6.5–6.8) to expose its positively charged histidine molecules and targeting ligands of glucosamine molecules. In contrast, MAA is protonated at a pH of 5.0; therefore, ML-HGD responded to pH changes not at the TME, but in intracellular endosomes/secondary lysosomes.
ML-HGD exhibit less pH-triggered IMQ release behavior (Fig. 3A). In contrast, the cumulative release percentage of IMQ from PL-HGD was significantly increased below a pH of 6.5 compared to that at a pH of 7.4 because the protonation of PAA and histidine makes the PL-HGD swollen (Fig. 3B). The experimental results indicate that PL-HGD could release IMQ at the TME for the activation of DCs and TAMs. The cumulative release percentage of IMQ at pH 6.5 was increased by 20% compared to pH 7.4, exhibiting statistically significant differences. This study has revealed that only 1 μg/mL of IMQ could cause proinflammatory cytokine secretion of bone-marrow-derived dendritic cells [38]. Therefore, the IMQ released from micelles at pH 6.5 was sufficient to induce the activation of DCs. On the other hand, DOX was conjugated onto the side chain of copolymers through hydrazone bonds; therefore, both ML-HGD and PL-HGD released low amounts of DOX in neutral and acidic surroundings (pH < 7) (Fig. 3C and D). The M-HGD and P-HGD copolymers were observed to aggregate in DMSO (Fig. S14), suggesting that the hydrophobicity of the copolymers was remarkably increased after conjugating with glucosamine, histidine, and DOX. This result suggests that the hydrophobicity of the copolymers was remarkably increased after conjugating with glucosamine, histidine, and DOX. Furthermore, drug release behavior has been reported to be associated with the hydrophobicity of nanoparticles [39]; therefore, it is challenging to hydrolyze the hydrazone bonds and release the DOX from the tight structure of core nanoparticles. However, the changes in the sizes of PL-HGD were more noticeable than those of ML-HGD, leading to a faster cumulative release of DOX from PL-HGD than from ML-HGD as pH decreases. Previous studies have shown that acid hydrolases in the secondary lysosomes might be helpful in facilitating hydrazone bond cleavage between the DOX and polymers [40,41]. Therefore, the relatively fast release of DOX from PL-HGD in cancer cells was expected. Glucose Transporter 1 (GLUT1) is overexpressed in various mammary carcinoma cells such as 4T1, MCF-7, and MDA-MB-231 [42,43]; therefore, glucose has been used as an active ligand of nanoparticles to target cancer cells [26,44]. An acidic TME is known to be closely associated with tumor chemosensitivity [45,46]. In a mildly acidic environment (such as TME), DOX, a common chemotherapeutic drug with a low base pKa value, becomes charged and increases the polarity leading to higher hydrophilicity; therefore, DOX can hardly pass through the cytoplasmic membranes, which reduces its cytotoxicity and therapeutic efficacy [47]. Here, in Fig. 3E and F, the cellular uptake of DOX significantly decreased (around 50% of the DOX mean fluorescence intensity (MFI)) as the pH level changed from 7.4 to 6.0. In contrast, the MFI for ML-HGD was almost the same in neutral and acidic surroundings. The cellular uptake of ML-HGD was lower than that of PL-HGD, probably because of the DOX quenching effect in nanoparticles and lower cumulative release rates in acidic surroundings. Moreover, MFIs of the PL-HGD at pH levels of 6.5 and 6.0 were 19.1 and 32.4%, respectively, which were higher than that at a pH of 7.4. The cellular uptake of PL-HGD at pH 6.0 was 1.32-fold higher than that at pH 7.4. After pretreatment with 100 mg/mL glucose at pH 6.0, the cellular uptake of PL-HGD decreased to 109%, indicating that the positive charges only promoted approximately 9% of cellular uptake (Fig. 3G). The MFI for PL-HGD was 2.5-fold higher than that for free DOX at pH 6.0, suggesting that PL-HGD could increase the intracellular accumulation of DOX because glucose ligands get exposed at low pH levels. In addition, the fluorescent images indicate that free DOX accumulated in the nucleus but separated with cytoplasm. The fluorescence intensity of the free DOX decreased with decreasing pH. In contrast, the DOX signal released from the ML-HGD was co-localized with the LysoTracker and slightly increased as pH decreased. However, the DOX released from the PL-HGD occupied both the nucleus and cytoplasm. The fluorescent intensity for the PL-HGD noticeably increased as the pH decreased and was much higher than that for the ML-HGD, indicating that the PL-HGD had targeted and rapid drug release abilities at low pH (Fig. 4). However, the MFI for PL-HGD at a pH of 6.0 was dramatically decreased in the glucose-pretreated competition test, suggesting the inhibition of GLUT-1-mediated internalization of PL-HGD. These cell uptake experiment results indicate that PL-HGD could expose the targeting ligand in the TME and improve DOX accumulation in cancer cells to overcome the low penetration of DOX in acidic surroundings.
In order to understand the cytotoxicity of the micelles on cancer and normal cells, 4T1 and L929 cells were co-incubated with DOX, IMQ, DOX combined with IMQ, and micelles at various concentrations. The experimental results reveal that IMQ was non-toxic to 4T1 cells because the concentrations of IMQ were too low to damage cells (Fig. 5A and B). Furthermore, DOX combined with IMQ exhibited similar cytotoxicity as DOX alone, indicating a non-synergistic effect at these concentrations. The cytotoxicity of PL-HGD treatment on 4T1 cells was similar to those of free DOX and DOX combined with IMQ, indicating that the DOX released from PL-HGD efficiently damages 4T1 cells. In contrast, the cytotoxicity of ML-HGD was lower than those of free DOX and PL-HGD because of lower DOX releasing ability. On the other hand, the cytotoxicity of ML-HGD and PL-HGD on L929 cells were much lower than that of DOX (Fig. 5C and D), suggesting that free DOX was toxic to normal cells, whereas micelles were not. In order to demonstrate that the cytotoxic effects of micelles were induced by DOX, the cytotoxicities of copolymers, without DOX conjugation, on 4T1 cells was evaluated. The M-HG and P-HG copolymers without DOX conjugation did not exhibit cytotoxicity even at a concentration of 1000 μg/mL (Fig. 5E and F). These cytotoxicity results indicate that micelles exhibited effective cancer cell cytotoxicity even though the release of DOX from micelles was slow. Cytotoxic effects of free DOX and micelles on RAW 264.7 cells were also evaluated by the MTT assay. After 24 h of incubation, free DOX exhibited a high cytotoxicity to RAW 264.7 cells (Fig. 6A). In contrast, both ML-HGD and PL-HGD induced the proliferation of RAW 264.7 cells at low DOX concentrations because of the regulatory effects of IMQ [48]. However, the cell viability dramatically decreased at high DOX concentrations for ML-HGD treatment because cytotoxic effect of DOX was stronger than the proliferation effect induced by IMQ. PL-HGD induced higher cell proliferation than ML-HGD because of the high release rate of IMQ at low pH levels. These cytotoxicity results showed that DOX is highly cytotoxic to cancer cells (4T1), fibroblasts (L929), and macrophages (RAW 264.7). In contrast, PL-HGD caused 4T1 cell death and were less cytotoxicity to L929 and RAW 264.7 cells. TLR7 is localized in phagosomes and can be used to sense foreign single-stranded RNA [49]; therefore, its activation can induce three downstream signaling pathways including mitogen-activated protein kinases (MAPKs), interferon regulatory factors (IRFs), and nuclear factor kappa–light-chain-enhancer of activated B cells (NF-κB). Eventually, proinflammatory cytokines including IL-6, IL-12, IL-18, and TNF-α can be produced by M1-like macrophages [50]. Our cytokine secretion results indicated that both ML-HGD and PL-HGD can significantly induce TNF-α and IL-6 secretion at low (0.1 μg/mL) and high (1.0 μg/mL) doses of IMQ; whereas, free IMQ only elevated the level of TNF-α (Fig. 6B). The cytokine secretion performance of PL-HGD was better than that of ML-HGD. The morphologies of RAW 264.7 cells were observed after treatment with free drugs and micelles. RAW 264.7 cells treated with DOX or DOX combined with IMQ had a spherical shape (dead cells) (Fig. 6C); whereas, RAW 264.7 cells had a spindle-shaped morphology (M1 type) after treatment with IMQ alone or micelles. CD86, a M1 type macrophage marker, was also evaluated at low (0.1 μg/mL) doses of IMQ and micelles. The expression level of CD86 on RAW 264.7 cells treated with PL-HGD was significantly increased among treating groups (Fig. 6D). These experimental results suggest that PL-HGD could polarize macrophages to M1 phenotype as compared to free IMQ and ML-HGD.
The tumor-bearing mice were intravenously injected with Cy5.5-labeled ML-HGD and PL-HGD to observe the distribution of the micelles in the organs and tumors. The tumor accumulation of PL-HGD was two times higher than ML-HGD at 24 h post-injection (Fig. 7A and B). In addition, DOX intensity from free DOX or micelles in the tumors were also evaluated after tumor-bearing mice were injected testing samples at a dose of 8 mg/kg equivalent of DOX for 24 h. The experimental results show that PL-HGD had a higher DOX signal compared to those of free DOX and ML-HGD (Fig. 7C and Fig. S15-19) because the exposed targeting ligands assisted the PL-HGD to accumulate in the tumors by preventing it from washing out of the tissues at the TME [51]. On the other hand, our previous study was showed that free DOX could accumulate in the liver and spleen at 24 h while the fluorescent intensity of DOX for those organs was low. At 72 h post-injection, an obvious fluorescent signal was observed in the liver and other organs [52]. Therefore, free DOX had a higher accumulation in the tumor than in other tissues at 24 h. Additionally, the hemocompatibility of micelles was assessed. Both ML-HGD and PL-HGD exhibited lower hemolysis than free DOX, indicating that our nanoparticles had high hemocompatibility (Fig. S20 and S21). PL-HGD had better stability, pH-responsiveness, cancer cell cytotoxicity, cytokine secretion, and tumor accumulation than ML-HGD. In addition, the concentration of encapsulated IMQ and conjugated DOX was difficult to adjusted to the same level for PL-HGD and ML-HGD. Therefore, PL-HGD was selected for further evaluation of antitumor activity. h. For the antitumor study, the 4T1 orthotopic tumor-bearing mice were randomly divided to 6 groups and injected intravenously with PBS, DOX, IMQ, DOX combined with IMQ, PL-HGD, and 2-fold PL-HGD on days 0, 3, 6, and 9. The results of tumor growth inhibition show that 2-fold PL-HGD significantly reduced tumor growth without weight loss (Fig. 8A and B). PL-HGD could also inhibit tumor growth in the initial 8 days as well as 2-fold PL-HGD. As time increased, there were no significant differences in the treatments of DOX, DOX combined with IMQ, and PL-HGD on day 12. However, mice treated with DOX or DOX plus IMQ showed severe body weight loss. The tumor volume after free IMQ treatment was not significantly different (P = 0.918) from the control group because the dose of IMQ administered via intravenous injection was too low to inhibit tumor growth in the 4T1 tumor-bearing mice [6]. In order to confirm the antitumor activity, tumor tissues from each group were stained with hematoxylin and eosin (H&E). Although the cell number was observed to be reduced in all treated groups, PL-HGD and 2-fold PL-HGD had better performance than other treatments (Fig. 8C). In contrast, DOX and DOX combined with IMQ induced higher levels of serum GPT and highly reduced WBC counts compared to those in the control group (Fig. 8D and E). The serum BUN in free DOX, DOX + IMQ and PL-HGD group were significant difference compared to control group. Actually, the level of GOT, BUN and CRE were still within the normal range (Fig. 8E and Fig. S22) [53,54]. The major organs after PL-HGD treatment did not exhibit pathological differences compared with the control group, indicating that PL-HGD was relatively safe (Fig. S23). Although PL-HGD exhibited similar antitumor activity with DOX and DOX combined with IMQ, weight loss and WBC reduction were not observed. These adverse effects are the major reasons for chemotherapy failure, limiting the use of the chemotherapeutic agents [55]. In this study, PL-HGD exhibited lower cytotoxic effects and similar therapeutic efficacy compared to those of DOX and DOX combined with IMQ. Although the antitumor efficacy of PL-HGD significantly improved upon increasing the dose up two-fold (2-fold PL-HGD), the mice did not exhibit any weight loss, high serum GPT levels, and WBC reduction. As the DCs sense cancer antigens mediated by IMQ, they can transform from their immature state to an activated form [6]. Activated DCs have been shown to express high levels of CD80, CD86, CD83, and CD40 on cellular membrane [56,57]. Reports have also shown that activated DCs can migrate to lymph nodes in order to present antigens to T cells through the T-cell receptors [58], In order to evaluate the immune status, mice were sacrificed after 12 days of treatments. The inguinal lymph nodes and tumors were collected, dissociated, and co-stained with specific marker. Furthermore, tumor tissue slices were also stained with anti-CD3 and anti-CD8, which are important biomarkers of cytotoxic T cells [59]. In this study, mice that received DOX combined with IMQ treatment exhibited lower CD80, CD86, and CD11C expressions compared to those of the control group. In contrast, PL-HGD and 2-fold PL-HGD were able to increase the percentages of CD80+, CD86+, and CD11C+ cells, indicating that they could facilitate the transformation of DCs into their mature form in lymph nodes (Fig. 9A). In addition, tumor tissue after 2-fold PL-HGD treatment had the highest M1/M2 ratio (around 6.46); the M1/M2 ratios for the control, free DOX + IMQ, PL-HGD and 2-fold PL-HGD were 0.12, 0.08, 1.04 and 6.45, respectively (Fig. 9B). The experimental result indicated 2-fold PL-HGD not only dramatically increased the percentages of F4/80+/CD86+ cells (M1 macrophages) but decreased the population of F4/80+/CD206+ cells (M2 macrophages). The population of CD3+/CD8+ cells increased after PL-HGD treatment (Fig. 9C); the IMQ induced upregulation of the costimulatory signal (CD80/CD86) on the surface of the dendritic cell to stimulate CD8+ T cell activation [[60], [61], [62]]. On the other hand, the immunohistochemical staining of the tumor sections revealed that both CD3+ and CD8+ T cells dramatically increased after treatments with PL-HGD and 2-fold PL-HGD (Fig. 10A), demonstrating that activated cytotoxic T cells recognized specific antigens presented by activated DCs and penetrated into the tumors. Upon treatment with PL-HGD and 2-fold PL-HGD, the expression of TNF-α and iNOS were notably increased (Fig. 10A,B and Fig. S24). Immune cells within the TME are crucial for an anti-cancer effect. Previous studies have shown that high densities of CD3+ and CD8+ T cells in the tumor interior and the invasive margin region indicate lower tumor recurrence [63]. Proinflammatory cytokines including TNF-α and iNOS have been shown as important markers of M1-like macrophages [64,65]. Polarization of macrophages from M2 to M1-type has been shown to contribute to antitumor immunity and inhibition of angiogenesis [66]. In addition, high M1/M2 ratio in tumor tissue exhibits better patients' survival time and prognosis [67]. Our antitumor efficacy results reveal that PL-HGD could deliver required dosages without causing adverse effects, inhibit tumor growth, and trigger immune responses for combined immunotherapy, which did not occur with DOX and IMQ treatments.
In this study, dual pH-sensitive and TME-active targeting micelles were designed for combined chemo- and immunotherapy to treat cancer. These micelles could exactly expose glucose ligands and positive charges to target cancer cells in the TME and release DOX into cancer cells for chemotherapy. Furthermore, these micelles also released IMQ inside TAMs and in the TME, thereby polarizing TAMs towards the M1-like phenotype for macrophage-mediated immunotherapy and maturing DCs for immunotherapy. Our results demonstrated that dual pH-sensitive and TME-active targeting micelles were selectively toxic to 4T1 cells without damaging normal cells and macrophages and exhibited high tumor accumulation, resulting in significant tumor growth inhibition and fewer adverse effects. Micelles could also induce high percentages of mature DCs in the lymph node and increase the densities of CD3+, CD8+ T cells, as well as M1-like macrophages in tumor tissues. Although treatment with these dual pH-sensitive and TME-active targeting micelles did not completely destroy all cancer cells, it demonstrated that precisely designing the structure of nanomedicines can achieve a therapeutic effect, validating this multi-treatment approach to treat cancer.
Yu-Han Wen: Conceptualization, Methodology, Investigation, Visualization, Writing - Original Draft, Writing - Review & Editing. Po–I Hsieh: Investigation, Visualization. Hsin-Cheng Chiu: Conceptualization, Supervision. Chil-Wei Chiang: Investigation. Chun-Liang Lo: Conceptualization, Visualization, Supervision, Project administration, Writing - Review & Editing. Yi-Ting Chiang: Conceptualization, Supervision, Resources.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. |
|
PMC9647597 | Neda Karami,Hadi Aligholi,Moosa Rahimi,Hassan Azari,Tahereh Kalantari | G Protein Coupled Receptors Potentially Involved in Oligodendrogenesis: A Gene Expression Analysis | 01-10-2022 | Oligodendrocyte progenitor cells (OPCs),Oligodendrocytes,Remyelination,Demyelination,Gene expression analysis | Background: Multiple sclerosis (MS) is a chronic demyelinating disease of the central nervous system characterized by infiltration of inflammatory leukocytes to the CNS followed by oligodendrocyte cell death, myelin sheath destruction, and axonal injury. A logical incidence occurring after demyelination is remyelination. G-protein coupled receptors (GPCRs) activate internal signal transduction cascades through binding to different ligands. This family of receptors are targeted by more than 40% of currently marketed drugs. GPCRs can be successfully targeted for induction of remyelination. GPCRs highly enriched in oligodendrocyte progenitor cells compared to oligodendrocytes are proposed to hamper oligodendrocyte differentiation and therefore their inhibition might induce remyelination. This study aimed to investigate the expression of GPCRs in silico and in vitro. Methods: We performed gene expression analysis using DAVID and Panther websites on a RNA-seq dataset (GSE52564 accession number). Primary embryonic neural stem/progenitor cell isolation and culture were performed and subsequently NSPCs were characterized by Immunocytochemistry with Anti-Nestin antibody. Expression of GPR37L1, EDNRB, PDGFRα, CNPase and GFAP were assessed using real-time PCR. All the experiments were conducted at Shiraz University of Medical Sciences (SUMS), Shiraz, Iran, in the year 2018. Results: The 14 most highly expressed GPCRs in oligodendrocyte progenitor cells (OPCs) compared to Oligodendrocytes were presented in our study. Conclusion: The investigation of the most highly expressed GPCRs in OPCs compared to oligodendrocyte in silico and in vitro presents the significant role of GPCRs in remyelination induction. Among the 14 GPCRs mentioned in this study, GPR37L1 is a potential remyelinating drug target and is suggested for further studies. | G Protein Coupled Receptors Potentially Involved in Oligodendrogenesis: A Gene Expression Analysis
Multiple sclerosis (MS) is a chronic demyelinating disease of the central nervous system characterized by infiltration of inflammatory leukocytes to the CNS followed by oligodendrocyte cell death, myelin sheath destruction, and axonal injury. A logical incidence occurring after demyelination is remyelination. G-protein coupled receptors (GPCRs) activate internal signal transduction cascades through binding to different ligands. This family of receptors are targeted by more than 40% of currently marketed drugs. GPCRs can be successfully targeted for induction of remyelination. GPCRs highly enriched in oligodendrocyte progenitor cells compared to oligodendrocytes are proposed to hamper oligodendrocyte differentiation and therefore their inhibition might induce remyelination. This study aimed to investigate the expression of GPCRs in silico and in vitro.
We performed gene expression analysis using DAVID and Panther websites on a RNA-seq dataset (GSE52564 accession number). Primary embryonic neural stem/progenitor cell isolation and culture were performed and subsequently NSPCs were characterized by Immunocytochemistry with Anti-Nestin antibody. Expression of GPR37L1, EDNRB, PDGFRα, CNPase and GFAP were assessed using real-time PCR. All the experiments were conducted at Shiraz University of Medical Sciences (SUMS), Shiraz, Iran, in the year 2018.
The 14 most highly expressed GPCRs in oligodendrocyte progenitor cells (OPCs) compared to Oligodendrocytes were presented in our study.
The investigation of the most highly expressed GPCRs in OPCs compared to oligodendrocyte in silico and in vitro presents the significant role of GPCRs in remyelination induction. Among the 14 GPCRs mentioned in this study, GPR37L1 is a potential remyelinating drug target and is suggested for further studies.
Oligodendrocytes are cells that make and sustain the lipid-rich myelin sheaths insulating and en-wrapping axons. During development, these cells arise from oligodendrocyte progenitor cells(OPCs) and neural stem cells (NSCs). In the Central Nervous System (CNS) an insult directed at the oligodendrocytes is the most preliminary cause of demyelination. Demyelination is a condition in which concentric layers of compact myelin, surrounding axons are lost (1). Defects in both oligodendrocyte development and demyeli-nation cause various neurological disorders, such as spinal cord injury and multiple sclerosis (MS) (2, 3). A logical corollary to CNS demyelination is frequently the robust regenerative process of remyelination. The source of remyelination is provided by newly formed oligodendrocytes in adult brain through differentiation of Neural stem/progenitor cells (NSPCs) or OPCs (4). Switching from OPCs differentiation to oligodendrocytes is mostly halted by various factors in different diseases. To overcome this incident and awaken the sequestered reservoir of remyelination, pharmacological activation of these cells has recently become the center of attention (5). There is the possibility of re-purposing the currently marketed drugs for remyelination. G-protein-coupled receptors (GPCRs) trigger one of the most common response pathways in the cell formed by the guanine nucleotide-binding proteins (G proteins). Approximately half of all drugs currently on the market target these receptors (6). Hunting for new drug targets can be tedious and costly through functional screening of drugs in 2D cell cultures for remyelination. A study published in Nature shed light on the efficacy of two drugs, miconazole and clobetasol in promoting precocious myelination in vivo in early postnatal mouse pups (7). A study discovered Benztropine (a well-established approved drug for the treatment of Parkinson’s disease) to be the most effective inducer of OPC differentiation (8). The targets for these proposed drugs are mostly GPCRs. Rational screening of drug targets using in silico tools is another faster and less costly way for drug target discovery. Studies for functional rational drug/drug target screening for remyelination were lacking. In the current study, we evaluated differential GPCRs expression in oligodendrocytes and OPCs to rationally propose drug targets for remyelination. In an RNA-seq data analysis using panther and David websites in the current study, we investigated the expression level of GPR37L1 a closely related orphan GPCR to GPR37 (9–11). In addition, we derived the most expressed GPCRs in oligodendrocyte progenitor cells compared to mature oligodendrocytes. Several of the found GPCRs have already been linked to remyelination induction. Among the GPCRS, GPR37L1 is a potential drug target for remyelination induction. GPR37L1 is a constitutive expressed orphan GPCR with a distinct expression pattern in glial cells and CNS. The paucity of information on this receptor has cast a shadow over its possible fascinating role in myelination. We proposed GPR37L1 and other mentioned GPCRs as drug targets for remyelination induction and promotion of oligodendrocyte differentiation.
RNA-seq data was downloaded from the website (https://www.stanford.edu/). The accession number from GEO datasets was GSE52564. The data sets were analyzed using DAVID and Panther websites. First, the differential gene analysis following annotation of RNA-seq data were performed. Next, we did a gene set enrichment analysis displaying different gene sets (such as GPCRs and lipid biosynthesis gene sets) upregulated in OPCs and OLs. Finally, we derived the pathway analysis in both cell types and compared the expression of highly expressed pathways in each cell type. All these were done according to the published protocols of each website in the journal of Nature protocols (9, 10). All the experiments were conducted at Shiraz University of Medical Sciences (SUMS), Shiraz, Iran, in the year 2018. Animals, Surgery and Dissection of mice cortices All mice examined in this study were obtained from Comparative and Experimental Medical Center of Shiraz University of Medical Sciences (SUMS). All animal procedures were conducted following protocols approved by the Animal Ethics and welfare committee of SUMS (no. IR. SUMS.REC.1396.S448). Female and male BALB/c mice were mated at 1:2 ratio (male: female) and female mice were checked for the following next 5 d until vaginal plaque was observed. On the 14th day of the mouse pregnancy, pregnant female mouse was deeply anesthetized with ketamine and xylazine and sacrificed. 14.5 d old mice embryos’ heads were dissected with fine scissors to be used further for isolation of NSCs. The cortex was subtly dissected with fine forceps and scissors under dissecting microscope and placed in ice-cold Phosphate Buffered Saline (PBS) solution containing 10% penicillin-streptomycin (Penstrep).
Following the dissection procedures described above, we pipetted the cell pellet in 1 ml warm Neurocult proliferation medium and counted the number of viable cells using trypan blue 0.4% dead cell exclusion method(12). 2 ×105 viable cells per 1 ml of complete neural stem cell culture medium are cultured in T25 flasks (SPL). For a T25 Flask we cultured the cells with complete medium containing Neurocult proliferation medium and supplement at a (1:9 ratio), 20 ng/ml Epidermal Growth Factor (EGF) and 1% penstrep. Cell cultures were incubated at 37 °C in 5% CO2 for 5 to 7 d and every 2 d, half of the medium was changed. This was considered passage 0. NSPCs from passages 2–4 were used for the experiments every 5 or 7 d, cells were passaged by tripsinization and then centrifuged at 1200 g for 5 minutes. The single cells were then seeded at 1×105 cells/ml in complete medium.
NSPCs were seeded at a density of 5000 cells/well in poly-l-ornithine coated plates (SPL) in complete neural stem cell proliferation for 7 days. The coating procedure was performed in a way to let neural stem cells form neurospheres; diluted (1:4) Poly-l-ornithine in PBS was incubated in 96 well plates for only less than an hour in an incubator. After 7 days wells were gently washed twice with PBS and fixed in 1% paraformaldehyde for 20 minutes. Fixed cells are then permeablized in 0.1% TritonX-100 for 5 min for nestin (a cytoplasmic marker). Nonspecific antibody binding sites were blocked by incubating with 5% Normal Goat Serum and 1% Bovine Serum Albumin in PBS for 1 h. Cells are rinsed twice with PBS and labeled with mouse nestin monoclonal antibody (1:200) in PBS containing 10% BSA at 4 °C overnight. Before staining the nuclei with 7-Aminoactinomycin D (7-AAD) nuclear stain for 1 min, wells are rinsed cautiously with PBS and then incubated with FITC-conjugated anti-mouse nestin antibody (1:600) for 2 h at room temperature. Immunofluorescent-labeled cells were visualized with Nikon Eclipse TS100 microscope coupled with a True Chrome Metrics camera. Images were taken at 10X magnification from 3 random areas in each well (n=3 for each condition). Images were processed using Photoshop CC 2016 and the intensity of each image was analyzed using ImageJ software.
The NSPCs were plated at a density of 2 × 105 Cells per 24 well plates coated with Poly-lornithine in complete neural stem cells media. Total RNA was extracted from cultured cells using RNXplus reagent (Cinnagen) according to the manufacturer’s instructions. Total RNA (1μM) was reverse transcribed into cDNA using Prime Script II First Strand cDNA synthesis Kit (TaKaRa), and 0.1% of cDNA mixture was used as polymerase chain reaction (PCR) template. Primers are shown in Table 1. The reaction was performed using a SYBR Green PCR Master Mix Kit (Yektataghiz azama) in a Rotor-gene Q (Qiagen) with an initial denaturation step at 95 °C for 30 sec, following 45 cycles. Each cycle transitioned between 3 steps of 95 °C for 5 sec, 60 °C for 30 sec and an extension step of 72 °C for 30 sec. Beta Actin was employed as the housekeeping gene to account for sample variability. Relative gene expression is represented as Fold change (2 −ΔΔCt).
For fluorescence intensity analysis and Percentage of antibody, positive cells calculation Image J software was used. The total number of cells and antibody-stained cells were counted manually and automatically by Image J software. For the neurosphere assay (estimating the diameter of the Spheres) Infinity Analyze version 4.6 was used. For primer Design Primer-BLAST and Gene runner 6.5.51 was used. Rotor-Gene Q software, ver. 2.3.1 was used to visualize and partially analyze the Real-time PCR results. Raw data were analyzed in Excel 2014.
All experiments were conducted at least in triplicate (n=3). Data were analyzed using Graphpad prims software (ver. 6.0); Data in the figures are expressed as mean ± SEM. Two-way ANOVA followed by Sidak multiple comparison post hoc was performed to compare differences among multiple treatments. Independent samples t-test was performed to compare differences between two conditions. For all experiments significance was defined as P-value<0.05.
In this study, we analyzed the gene expression profiles of 5 different cells. The comparison of gene expression showed a high resemblance in the expression pattern of Gpr17, EDNRB, Gpr56, Gpr19 and GPR37L1 as indicated in Fig. 1. However, EDNRB and GPR37L1 expression patterns are more alike, both being upregulated in OPCs and astrocytes and lower expression in newly myelinating oligodendrocytes. Whereas, their expression in myelinating oligodendrocytes compared to OPCs is merely unnoticeable. Whether GPR37L1 shares a similar function with EDNRB on oligodendrocytes differentiation is quite an interesting dilemma (13, 14). The expression of Chrm1, Chrm2, Hrh1 and P2ry1 was not significant. Gpr37 and Gpr62 have the same expression pattern too but it was not significant to be discussed.
Primary NSPCs culture formed neurospheres of > 200 μM in diameter (P0) after 7 d of culture. The expression of Nestin decreased (data not shown) and cell culture time was shortened to ∼ 5 d in the following passages (Fig. 2). The percentage of PDGFR-alpha mRNA expression didn’t show a significant increase in passage 4 compared to passage 2 of cortical neural stem cells culture, albeit CNPase showed a noticeable fold change expression of 1.66 and GFAP mRNA expression was downregulated by almost 5 fold change (50%). PDGFR-alpha is a marker for oligodendrocyte progenitor cells, CNPase is upregulated in mature pre-myelinating and myelinating oligodendrocyte, expressed along MBP and PLP constituents of myelin sheath. As commonly known GFAP is an astrocyte marker, but is also abundantly expressed in radial glial cell (which is the progenitor cell giving rise to both OPCs and astroglial cells) (Fig. 2).
Eendothelin B and GPR37L1 expression exhibited greatly increased expression during further Cortical NSPCs passages (with EDNRB and GPR37L1 gene expressions enhanced by almost 3 and 6 fold change in cortices derived from passage 4 compared to passage 2) as shown in Fig. 2. EDNRB and GPR37L1 expression were explored in NSPCs derived from embryonic cortices compared to ganglionic eminences (both at passage 02) alongside Glial markers (CNPase and PDGFR-alpha) and astrocytic marker GFAP. Both EDNRB and GPR37L1 displayed higher gene expressions in NSPCs derived from cortices compared to ganglionic eminences.
In this study, we presented the possible role of GPCRs in driving remyelination. We conducted gene expression analysis on a RNA-seq dataset of genes expressed in cells derived from mouse nervous system. The expression of GPCRs highly expressed in OPCs compared to oligodendrocytes are presented here. Most of these genes were previously reported to harm remyelination. Among these GPCRS, expression of EDNRB and GPR37L1 were investigated in vitro as well. Anti-Nestin staining was performed for characterizing NSPCs. Commonly Nestin has been utilized as a biological marker to identify NSCs (15). Cells express nestin early in their life cycle. Nestin expression is down-regulated as the cells progress down a specific cell lineage to become either neuron or glial cells (16, 17). NSPCs express nestin and further passages of cortical NSPCs show a slight reduction in Nestin expression (data not shown). The expression of Gpr37l1and Ednrb was explored at mRNA level using real-time PCR in conseqeutive passages of cortical NSPCs compared to Ganglionic eminences derived NSPCs. Our finding shows Gpr37l1 has a similar pattern of expression in mice compared to Ednrb. Developmentally, oligodendrocytes arise from OPCs (18, 19). OPCs themselves arise from subventricular cells in the brain and spinal cord. A myriad of different permissive and inhibitory factors orchestrate the differentiation of oligodendrocytes. Several Inhibitory factors are expressed by axons to usher myelination and differentiation of OPCs. Induction of remyelination can be addressed by activation of endogenous OPCs present around demyelinated lesions (2, 3, 20, 21). Our premise was that GPCRs highly expressed in OPCs compared to oligodendrocytes cast an inhibitory effect on oligodendrocyte differentiation. The most highly expressed GPCR genes in OPCs compared to mature oligodendrocytes were Gpr37l1, Ednrb, Gpr17, Gpr37 and Gpr56. Ednrb, Gpr17 and Gpr37 roles have all been investigated in the context of remyelination. Ednrb Regulates the Rate of Oligodendrocyte Regeneration during Remyelination. Other GPCRs such as Chrm1, Chrm2 and Chrna4 receptors belong to the family of muscarinic and cholinergic receptors and they all demonstrate a significant and mostly impermissive role in oligodendrocyte differentiation and remyelination. GPR17 is a P2Y purinergic GPCR affecting oligodendrocyte differentiation and myelination. GPR17 cast a negative effect on this phenomenon. The absence of Gpr17 enhances remyelination and the activation of Erk1/2 pathway following Gpr17 down-regulations corroborate this finding (22, 23). An article released in 2009 portrayed the role of Gpr17 in remyelination, downregulated in Olig1-null mice(24). Recently identification of a non-specific antagonist called pranlukast accelerated myelination following toxin-mediated demyelination (25). Endothelin B receptor was recognized as a potential inhibitory drug target that works in a paracrine and autocrine way to inhibit OPCs differentiation. ETBR is expressed on both astrocytes and OPCs(26). Coupling of ET-1 ligand to this receptor on astrocytes promotes Notch activation in OPCs during remyelination through induction of Jagged 1 expression in reactive astrocytes (13, 14). In a microarray-based experiment of isolated cells, GPR37 was shown to be strongly enriched in mature and pre-mature oligodendrocytes (27). GPR37 mutant mouse exhibits premature oligodendrocyte differentiation, precocious myelination and hyper myelination (28). The mechanism by which GPR37 regulates multiple stages of myelination is elusive. Nevertheless, given its strong enrichment in oligodendrocyte lineage, it can be an interesting drug target. GPR37 is structurally closely related to endothelin B receptor. GPR37 is an orphan GPCR distinctly expressed in neuronal and glial cells of the CNS. A negative regulatory effect of GPR37 was manifested on oligodendrocyte differentiation and myelination (29–31). GPR37 and Gpr37l1 act as parkin substrates. They are expressed in different CNS areas. The absence of these receptors caused an increase in ERK1/2 phosphorylation in both cultured oligodendrocytes and leads to decreasing myelin growth (29, 32). Moreover, lacking GPR37 showed changes in the expression of oligodendroglial proteins such as myelin associated glycoprotein (MAG) (33). A highly desirable property of a druggable target is its tissue or cell-type specific expression, reducing the concern over unwanted effects. GPR37L1 is an orphan GPCR exclusively expressed in the nervous system and are known to be expressed on both neurons and glial cells (30). Another closely related GPR37L1 is called GPR37 or also known as parkin associated endothelin-like receptor or “Pael-R”. the suggested cognate ligand for these two receptors is prosaposin which is still under investigation. Prosaposisn and prosaptide have been numerously reported to exert neuroprotective and oligo-protective effects. Through a bioinformatics approach, several surrogate ligands were proposed to inhibit GPR37l1, one of them is an orexin 2 receptor antagonist called JNJ10397049 (34).
The investigation of the most highly expressed GPCRs in OPCs compared to oligodendrocyte in silico and in vitro presents the significant role of GPCRs in remyelination induction. Among the 14 GPCRs mentioned in this study, GPR37L1 is a potential remyelinating drug target and is suggested for further studies. Other GPCRs presented in this study (Gpr56, Gpr62, Gpr19, Gpr162 and Hrh1) have a probable role in remyelination induction as well and are presented for further analysis and experimentation in vitro and in vivo.
Ethical issues (Including plagiarism, informed consent, misconduct, data fabrication and/or falsification, double publication and/or submission, redundancy, etc.) have been completely observed by the authors. |
|
PMC9647612 | Dong-Il Kim,Young-Min Park | Effects of Menopause on Physical Activity and Dopamine Signaling in Women | 01-10-2022 | Effects of Menopause on Physical Activity and Dopamine Signaling in Women
More than 60% of women fail to engage in the U.S. guidelines of physical activity and a positive correlation has been found between physical inactivity and mortality in women. Women 60 yr of age and older show a greater prevalence of physical inactivity (1). Pre-clinical studies with ovariectomized rodents suggest that a lack of ovarian hormones reduces physical activity levels. A human longitudinal study (2) also demonstrated that women exhibited a significant reduction in physical activity 2 yr before menopause and remained reduced. This physical inactivity is associated with decreased circulating estrogen levels during the menopausal transition. These preclinical and clinical studies strongly suggested the existence of a significant physiological factor during menopause, and estrogen has been believed the main controller involved in menopause-related behavioral changes.
Mesolimbic dopamine circuits in the brain play a critical role in the regulation of motivation, motor control, and reward, which can significantly contribute to voluntary physical activity. Among the circuits, the dopamine system in the nucleus accumbens (NAc) seems to be a major controller for voluntary physical activity (3). There are two types of dopamine receptors in dopamine neurons, the D1-like receptor (stimulatory receptors, D1 and 5) containing no introns and the D2-like receptor (inhibitory receptors, D2, 3, and 4) containing introns acting through Gi-proteins (4). In a preclinical study, injecting a D1 receptor agonist into the NAc increases physical activity, while a D1 receptor antagonist decreases it (5). These findings confirmed the significant role of dopamine activity in voluntary physical activity. Estrogenic activation is the critical link for physical activity, and estrogen modulates neurotransmitters including dopamine (6). Menopause-related deficiency in estrogen decreases voluntary physical activity along with attenuated dopamine activity (7). Estrogen appears to exert a tonic stimulation for dopamine receptors and in turn maintain overall dopamine activity, which positive stimulation is attenuated after menopause in women or ovariectomized rodents. However, estrogen replacement therapy immediately after ovariectomy conserved the upregulated D1-like dopamine receptors (8). Our group previously determined the potential role of NAc dopamine activity in voluntary wheel running in female rats (9). Using rats selectively bred for high (HCR) and low (LCR) aerobic capacity showing a divergence in wheel running behavior, we found that HCR rats had greater wheel running distance and the activation of dopamine signaling compared to LCR rats. HCR rats had greater D1 stimulatory receptors and lower D3 & 4 inhibitory receptor mRNA expressions compared to LCR rats. However, ovariectomy significantly up-regulated inhibitory dopamine receptors (i.e. D2 and 4 receptor mRNA expressions) in HCR rats, implying a strong effect of menopause independent of one’s aerobic capacity. All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Missouri-Columbia.
ERα signaling is an obligatory mediator for the ovariectomy-mediated reduction in voluntary physical activity. Using genetically knock-out (KO) mice, Ogawa et al (10) initially demonstrated that ERα signaling appears more likely involved in enhanced spontaneous physical activity in ovariectomized mice, rather than ERβ-pathway. This group showed that estrogen administration increased spontaneous physical activity only in ERβ KO mice but failed in ERα KO mice. This strongly supports the hypothesis that the estrogenic increase in spontaneous physical activity is primarily mediated by the ERα-signaling pathway. To our knowledge, none of the previous studies investigated the effect of ERα signaling on dopamine receptors and voluntary running activity. Future studies should further investigate the specific mechanisms by which estrogen and its receptors (ERα and β) regulate brain dopamine metabolism in menopausal women. |
|||
PMC9647616 | Shicun Zhang,Fang Wu,Lan Zhan,Wennan Lin,Chunming Liang,Yu Pang,Jiawen Zhang,Zhuang Mu | Exercise Regulates the Lactate Receptor HCAR1 and ERK1/2-PI3K/Akt Pathways to Promote Cerebral Angiogenesis | 01-10-2022 | HCAR1 protein,Fibroblast-like cells,Angiogenesis | Background: We aimed to explore the ole and mechanism of lactate receptor (HCAR1) in the angiogenesis of leptomeningeal fibroblast-like cells. Methods: Human brain fibroblast-like cells were selected and some cells were deactivated, analyzed and compared with HCAR1 mRNA and protein expressions in deactivated/normal cells. HCAR1−/− mice and wild type (WT) mice were selected and divided into WT, WT exercise, HCAE1 KO and HCAE1 KO exercise groups, with 10 mice for each group. HCAR1mRNA and expression levels of proteins in fibroblast-like cells, mRNA and expression levels of proteins in Collagen IV, phosphatidylinositol trihydroxykinase (PI3K), serine threonine kinase (AKT) and extracellular signal-regulated kinases 1 and 2 (ERK1/2) in hippocampus were compared, and the microvessel density (MVD) and diameter were calculated. Results: mRNA and expression levels of proteins in Collagen IV, PI3K, AKT, ERK1/2 and MVD in hippocampus were significantly higher in the WT exercise group than those in the WT group, microvessel diameter was significantly lower than that in the WT group (P<0.05). mRNA and expression levels of proteins in Collagen IV, PI3K, AKT, ERK1/2 and MVD in hippocampus in the HCAR1 KO and HCAR1 KO exercise groups were significantly lower than those in the WT group, microvessel diameter was higher than that in the WT group (P<0.05). Compared with the HCAR1 KO exercise group, the changes of mRNA in Collagen IV, PI3K, AKT, ERK1/2 and microvascular were not significant. Conclusion: Exercise can promote cerebral angiogenesis through the activation of the lactate receptor HCAR1 and the ERK1/2-PI3K/Akt signaling pathways. | Exercise Regulates the Lactate Receptor HCAR1 and ERK1/2-PI3K/Akt Pathways to Promote Cerebral Angiogenesis
We aimed to explore the ole and mechanism of lactate receptor (HCAR1) in the angiogenesis of leptomeningeal fibroblast-like cells.
Human brain fibroblast-like cells were selected and some cells were deactivated, analyzed and compared with HCAR1 mRNA and protein expressions in deactivated/normal cells. HCAR1−/− mice and wild type (WT) mice were selected and divided into WT, WT exercise, HCAE1 KO and HCAE1 KO exercise groups, with 10 mice for each group. HCAR1mRNA and expression levels of proteins in fibroblast-like cells, mRNA and expression levels of proteins in Collagen IV, phosphatidylinositol trihydroxykinase (PI3K), serine threonine kinase (AKT) and extracellular signal-regulated kinases 1 and 2 (ERK1/2) in hippocampus were compared, and the microvessel density (MVD) and diameter were calculated.
mRNA and expression levels of proteins in Collagen IV, PI3K, AKT, ERK1/2 and MVD in hippocampus were significantly higher in the WT exercise group than those in the WT group, microvessel diameter was significantly lower than that in the WT group (P<0.05). mRNA and expression levels of proteins in Collagen IV, PI3K, AKT, ERK1/2 and MVD in hippocampus in the HCAR1 KO and HCAR1 KO exercise groups were significantly lower than those in the WT group, microvessel diameter was higher than that in the WT group (P<0.05). Compared with the HCAR1 KO exercise group, the changes of mRNA in Collagen IV, PI3K, AKT, ERK1/2 and microvascular were not significant.
Exercise can promote cerebral angiogenesis through the activation of the lactate receptor HCAR1 and the ERK1/2-PI3K/Akt signaling pathways.
The physiological role of lactate has been controversial after its discovery in biological tissues. Lactate is only an anaerobic metabolic waste, and with the deepening of its research, it can participate in the regulation of intracellular environmental homeostasis, and can also act as a signaling molecule (1,2). In recent years, lactate can be involved in the regulation of physiological cerebral functions and exert brain protective effects. For example, Roosterman et al (3) found that if the content of lactate in brain increases, the content of insulin-like growth factor, brain-derived neurotrophic factor, vascular endothelial growth factor (VEGF) will also increase, but the specific mechanism has not been clarified. Hydroxy-carboxylic acid receptor 1 (HCAR1), as a lactate-specific receptor, can be expressed in a variety of tissues and organs, such as adipose tissue, brain, gastrointestinal tract, etc. (4). In addition, HCRA1 expression significantly increases in tumor cells, indicating that it can participate in the regulation of tumor cells’ growth and development (5). HCAR1 plays an important role in the conduction of lactate and brain signaling, which can bind to lactate to reduce the spontaneous calcium peak frequency in brain, and is beneficial to blocking neuronal network activity (6,7). HCAR1 can bind to lactate to play a neuroprotective role, inducing VEGFA secretion and promoting cerebrovascular production, which plays a significant role in the repair of brain injury (8). By establishing the middle cerebral artery occlusion model, HCAR1 expression could be significantly up-regulated in both hippocampus and cerebral cortex, which could effectively improve the cerebral infarction area of middle cerebral artery occlusion model and improve the neuronal cell survival rate after the intervention of D-lactate and 3,5-DHBA (9). With the large number of brain fibroblasts-like cells, their functional status changes and their morphological structure changes (10). HCAR1 is highly enriched in leptomeningeal fibroblasts-like cells, indicating that HCAR1 may be a target for promoting cerebrovascular formation and enhancing brain function. However, the role of HCAR1 in fibroblast-like cells in the leptomeninges has not been fully elucidated, so we aimed to explore it to provide a theoretical basis for the clinical prevention and treatment of brain injury or neurodegenerative diseases.
Human brain fibroblast-like cells (Lot No.: HY-iCell-n004, from Beijing Vital River Laboratory Animal Technology Co., Ltd.) were taken for conventional culture and passaged at a ratio of 1:3 according to their proliferation rate to be studied in the logarithmic growth period.
A total of 20 2-month HCAR1+/− mice (Lot No.: SCXK (J) 2021-0006, purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd.) were fed in the same cage at a ratio between males and females of 1:1. This study was approved by the Animal Ethical Care Committee of Qiqihar Medical University (no. QMU-AECC-2021-225). The 1-month progenies were selected for gene phenotype identification. When the progenies were 2 months old, 20 HCAR1−/− mice and 20 WT mice weighed 18–20g were selected and fed in the SPF laboratory, with the temperature of (22±2) °C, the humidity of (55±5) %, and the replacement of day and night every 12h (Lighting event 08:30–20:30). Mice were divided into WT, WT exercise, HCAE1 KO, and HCAE1 KO exercise groups, with 10 mice for each group.
Human brain fibroblast-like cells were taken and placed in the 65°C water and killed in the ice water after 1min, while equal amount of live human brain fibroblast-like cells were added as positive controls.
Mice in WT exercise and HCAE1 KO exercise groups were trained with high-intensity exercise, placed on the treadmill and exercised at uniform speed, gradually accelerated to 14 m per minute during the first 3min, and then continued exercise at that speed for 1h, 1 time daily for 5w. Mice in the WT and HCAE1 KO groups did not do any exercise. After 5w, each group of mice were anesthetized with isoflurane and killed with hippocampus taken along with part of tissues taken for isolation of fibroblast-like cells.
mRNA expression of HCAR1, Collagen IV, PI3K, AKT and ERK1/2 were detected by RTPC. Total RNA was extracted from human brain fibroblast-like cells or mouse hippocampus by the TRIzol reagent. Primer sequences were designed and reverse transcribed by cDNA by the reverse recording kit, corresponding gene transcription was detected by RT-PCR with β-actin as the internal reference and the relative expression was calculated with 2−ΔΔCt.
The expression of proteins in HCAR1, Collagen IV, PI3K, AKT and ERK1/2 was determined by Western blot. Human fibroblast-like cells or lysates for mouse hippocampus were collected and centrifuged at 12,000 RPM at 4 °C for 10 min. The concentration of proteins was determined by the BCA protein assay kit. Electrophoresis was performed according to the time determined by the protein marker. The protein was transferred to PVDF membrane, which was placed in 5% skim milk for sealing. After incubation with primary and secondary antibodies, the membrane was washed, and ECL was added for the film’s exposure, development and fixing, and the expression levels of target proteins in HCAR1, Collagen IV, PI3K, AKT and ERK1/2 were analyzed.
Sections of mouse hippocampus were prepared, with the thickness of 4∼5μm and images on sections of mice in WT, WT exercise, HCAE1 KO and HCAE1 KO exercise groups were collected by the high-resolution fluorescence image system and processed with the image processing software, such as Zen Lite Blue software. The SimpleGrid Plug-in for Image J system was selected to calculate the number and diameter of microvessels per 1mm2 according to the Delesse principle.
Data analysis was performed with SPSS 22.0 software (IBM Corp., Armonk, NY, USA), measurement data with (x̄ ± s), ANOVA for multiple group comparisons and LSD-t test for pairwise comparisons. P<0.05 was considered as a statistically significant difference.
HCAR1 expression was found in both normal and inactive brain fibroblast-like cells, and both HCAR1 mRNA and protein expressions were significantly higher in normal brain fibroblasts-like cells than those in inactive brain fibroblast-like cells (P<0.05, Fig. 1ABC). HCAR1 expression was not found in the HCAR1 KO group, but was found in the WT group (P<0.01, Fig. 1DEF). mRNA expressions of PI3K, AKT, ERK1 and ERK1/2 in hippocampus of mice in the HCAR1 KO group were (0.75±0.05), (0.68±0.08) and (0.55±0.04), respectively, which were (1.67±0.12), (1.49±0.20) and (1.84±0.17) in hippocampus of mice in the WT group. mRNA expressions of PI3K, AKT, ERK1 and ERK1/2 in hippocampus of mice in the HCAR1 KO group were significantly lower than those in the WT group (P<0.05). The protein expressions of PI3K, AKT and ERK1/2 in hippocampus of mice in the HCAR1 KO group were (0.78±0.06), (0.65±0.06) and (0.51±0.03), respectively, which were (1.68±0.12), (1.45±0.09) and (1.78±0.12) in hippocampus of mice in the WT group. Protein expressions of PI3K, AKT and ERK1/2 in hippocampus of mice in the HCAR1 KO group were significantly lower than those in the WT group (P<0.01), as shown in Fig. 2 and 3. mRNA and protein expressions of Collagen IV, PI3K, AKT and ERK1/2 in hippocampus of mice in the WT exercise group were significantly higher than those in the WT group (P<0.05); mRNA and protein expressions of Collagen IV, PI3K, AKT and ERK1/2 in hippocampus of mice in the HCAR1 KO and HCAR1 KO exercise groups were significantly lower than those in the WT group (P<0.01). However, there was no statistically significant difference in mRNA and protein expressions of Collagen IV, PI3K, AKT and ERK1/2 in hippocampus of mice of HCAR1 KO and HCAR1 KO exercise groups (P>0.05), as shown Fig. 4 and 5. MVD in the WT exercise group was significantly higher than that in the WT group, with the microvessel diameter significantly lower than that in the WT group (P<0.05). MVDs in the HCAR1 KO and HCAR1 KO exercise groups were significantly lower than that in the WT group, with the microvessel diameter significantly higher than that in the WT group (P<0.05). There was no statistically significant difference in MVD and microvessel diameter between HCAR1 KO and HCAR1 KO exercise groups, as shown Fig. 6.
Lactate is one of the energy sources of the central nervous system, which was not only as a metabolic fuel and buffer, but also as a signaling molecule involved in many pathophysiological links (11,12). HCAR1 plays an important role in regulating the energy metabolism of brain and blood flow in it (13,14). In the central nervous system, mRNA and proteins in HCAR1 are mainly expressed in hippocampus, neocortex and cerebellum. In addition, lactate signals transmitted by specific receptor proteins in the brain was found in the same study, with a preliminary analysis on lactate receptors step by step (15). We aimed to explore the role and mechanism of HCAR1 in the angiogenesis of leptomeningeal fibroblast-like cells, to provide theoretical support for the clinical prevention and treatment of brain injury or neurodegenerative diseases. HCAR1 is enriched in fibroblast-like cells in leptomeninges, indicating that brain fibroblast-like cells may become target sites to induce cerebro angiogenesis and improve cerebral functions, and HCAR1 may be an important target to induce cerebro angiogenesis (16). To further clarify HCAR1 expression in human brain fibroblast-like cells, we inactivated cells and detected HCAR1 expression in normal human brain fibroblast-like cells with RT-PCR and Western blot, respectively, which showed that mRNA and protein expressions of HCAR1 in normal brain fibroblast-like cells were significantly higher than those in inactivated brain fibroblast-like cells, suggesting that HCAR1 expressed both in normal and inactive brain fibroblast-like cells. The lactate/HCAR1 signaling systems can significantly attenuate neuronal excitability, and its mechanism may be associated with lactate activation of HCAR1 and AC/cAMP/PKA pathways (17). Lactate/HCAR1 signaling systems can play a neuroprotective role. It blocks the production of NLRP3 and the secretion of IL-1 in microglia, and can play a significant role in improving the neuroinflammatory response (18). Besides, lactate/HCAR1 signaling systems can activate ERK1/2 and AKT pathways, which is beneficial to induce VEGF expression and angiogenesis in hippocampus (19,20). Lactate/HCAR1 signaling systems can mediate different pathways to play neuroprotective and proangiogenic effects, and play an important role in the onset and evolution of neurological diseases. PI3K/AKT and ERK1/2 signaling pathways are important signaling systems in normal cells. Among them, PI3K/AKT is a common pathway for many membrane receptor signaling to intra-cellular cells. As one of the downstream effector molecules of PI3K, AKT is the key to this pathway, which can participate in cell biological behavior by regulating relevant proteins downstream of the pathway (21,22). ERK1/2 signaling pathway can be involved in the regulation of cell biological behavior and be activated by many extracellular signals, such as cytokines, hormones and neurotransmitters, but it has not been fully revealed whether targeting this pathway plays a role in brain injury or neurodegenerative diseases (23,24). In this study, we obtained HCAR1 knock-out mice and WT mice, took the hippocampus in mice and isolated the brain fibroblast-like cells, and then detected HCAR1 expression in the cells. It showed that no HCAR1 expression was found in brain fibroblast-like cells of HCAR1 knock-out mice, while HCAR1 expression existed in the WT mouse cells. In addition, we also tested the relevant indicators in hippocampus of mice, showing that mRNA and protein expressions of PI3K, AKT and ERK1/2 in the HCAR1 KO group were significantly lower than those in the WT group, which suggests that knocking down HCAR1 gene can block the activation of ERK1/2-PI3K/Akt signaling pathways. Repeated exercise can increase the content of lactate in brain to some extent, which is conducive to enhancing the cognitive ability (25). During the acute stress, elevated epinephrine can induce increased intracellular cAMP levels and promote increased cognitive function, however, under a chronic stress, a chronic increase of cAMP content can cause cognitive dysfunction (26). With increasing age, cAMP levels in human frontal cortices also increase, which then causes decreased cognitive function. At this time, repeated physical exercise can promote HCAR1 activation, and then improve the injury caused by the chronic increase in cAMP (27). Lactate is more active in skeletal muscle, and high intensity exercise can promote lactate accumulation in skeletal muscle in the blood, which is similar to lactate injection in brain. It can lead to the up-regulation of VEGFA expression in brain, thus induce neovascularization, which is beneficial to improving neurogenesis and synaptic functions, but the molecular signals related to the up-regulation of VEGFA expression in brain induced by high-intensity exercise have not been fully clarified (28). HCAR1 could be significantly activated by exercise and enhance the proangiogenic effect of lactate. Therefore, this study also used exercise mode to activate HCAR1, which was divided into WT group, WT exercise group, HCAE1 KO group and HCAE1 KO exercise group according to the mice’s intensity of exercise, aiming at exploring the role and mechanism of exercise and HCAR1 and PI3K/AKT/ERK1/2 signaling in vasculature formation. RT-PCR and Western blot showed that Collagen IV, C mRNA and protein expressions in hippocampus of WT mice after exercise intervention were significantly higher than those of WT mice without exercise. After knocking down of HCAR1, mRNA and protein expressions of the above indicators in hippocampus of mice were significantly lower than those of WT mice without exercise regardless of exercise intervention. MVD of angiogenesis can directly and clearly indicate the number of neovascular vessels and increased MVD, indicating that the degree of brain injury repair is ideal (29,30). In this study, we found that MVD in WT mice after exercise intervention was significantly higher than that in WT mice without exercise, and microvessel diameter was significantly lower than that in WT mice without exercise. After knocking out HCAR1 in mice, regardless of exercise intervention, MVD in hippocampus of mice was significantly lower than that in WT mice without exercise, and microvessel diameter was significantly higher than that in WT mice without exercise. It suggests that exercise may promote microangiogenic capillary angiogenesis in brain tissues by activating HCAR1 and ERK1/2-PI3K/Akt signaling pathways. After knocking out HCAR1 gene followed by high-intensity exercise, ERK1/2-PI3K/Akt signaling pathways in hippocampus of mice were activated and inhibited, and the expression of Collagen IV, PI3K, AKT and ERK1/2 decreased, and angiogenesis was blocked.
Exercise can promote microangiogenic angiogenesis in brain tissues. The mechanism may be realized by activating HCAR1 and ERK1/2-PI3K/Akt signaling pathways, to provide an experimental basis for mechanical research on brain injury or neurodegenerative diseases and new ideas for the treatment of cerebrovascular diseases.
Ethical issues (Including plagiarism, informed consent, misconduct, data fabrication and/or falsification, double publication and/or submission, redundancy, etc.) have been completely observed by the authors. |
|
PMC9647623 | Peng Wei,Zhifeng Dong,Ming Lou | Lncrna FGD5-AS1 Aggravates Myocardial Ischemia-Reperfusion Injury by Sponging Mir-129-5p | 01-10-2022 | Myocardial ischemia,Reperfusion injury,Cardiology | Background: LncRNA FGD5-AS1 regulates the pathogenesis of many human diseases. We aimed to elucidate the function of lncRNA FGD5-AS1 and the regulatory mechanism of lncRNA FGD5-AS1/miR-129-5p in myocardial ischemia-reperfusion (I/R) injury. Methods: Myocardial I/R injury mice model and H/R treated H9c2 cells were established. RT-qPCR and Western blot analysis were used to detect the mRNA and protein expression. Cell viability was detected by MTT assay. Dual luciferase reporter assay was applied to confirm the relationship between lncRNA FGD5-AS1 and miR-129-5p. Results: LncRNA FGD5-AS1 was upregulated in myocardial I/R injury mice models and H/R treated H9c2 cells. Functionally, knockdown of lncRNA FGD5-AS1 promoted cell viability and inhibited apoptosis in H/R treated H9c2 cells. In addition, lncRNA FGD5-AS1 directly targets miR-129-5p. Upregulation of lncRNA FGD5-AS1 weakened the protective effect of miR-129-5p on myocardial I/R injury. Conclusion: LncRNA FGD5-AS1 aggravates myocardial I/R injury by downregulating miR-129-5p. | Lncrna FGD5-AS1 Aggravates Myocardial Ischemia-Reperfusion Injury by Sponging Mir-129-5p
LncRNA FGD5-AS1 regulates the pathogenesis of many human diseases. We aimed to elucidate the function of lncRNA FGD5-AS1 and the regulatory mechanism of lncRNA FGD5-AS1/miR-129-5p in myocardial ischemia-reperfusion (I/R) injury.
Myocardial I/R injury mice model and H/R treated H9c2 cells were established. RT-qPCR and Western blot analysis were used to detect the mRNA and protein expression. Cell viability was detected by MTT assay. Dual luciferase reporter assay was applied to confirm the relationship between lncRNA FGD5-AS1 and miR-129-5p.
LncRNA FGD5-AS1 was upregulated in myocardial I/R injury mice models and H/R treated H9c2 cells. Functionally, knockdown of lncRNA FGD5-AS1 promoted cell viability and inhibited apoptosis in H/R treated H9c2 cells. In addition, lncRNA FGD5-AS1 directly targets miR-129-5p. Upregulation of lncRNA FGD5-AS1 weakened the protective effect of miR-129-5p on myocardial I/R injury.
LncRNA FGD5-AS1 aggravates myocardial I/R injury by downregulating miR-129-5p.
Myocardial ischemia refers to a pathological state in which the blood perfusion of the heart is reduced. It can lead to a decrease in oxygen supply to the heart and abnormal myocardial energy metabolism (1). With the improvement of people's living standards, the incidence of myocardial ischemia in China is increasing (2). Myocardial ischemia-reperfusion (M-I/R) injury indicates a cardiovascular dysfunction after myocardial ischemia (3). Currently, effective treatment is limited to restoring coronary blood flow to prevent myocardial infarction. Therefore, there is an urgent need for new therapeutic strategies to prevent M-I/R injury. Long non-coding RNA (lncRNA) is a type of RNA molecule with a transcript length of more than 200 nt. LncRNAs do not encode proteins, but regulate gene expression at multiple levels (epigenetic, transcription, post-transcriptional regulation) (4). Moreover, lncRNAs play important roles in the pathogenesis of M-I/R injury. For example, knockdown of lncRNA TTTY15 alleviated M-I/R injury through the miR-374a-5p/FOXO1 axis (5). In addition, lncRNA A2M-AS1 has been reported to lessen the injury of cardiomyocytes caused by hypoxia and reoxygenation via regulating IL1R2 (6). Here, the role of lncRNA FGD5-AS1 was investigated in the pathogenesis of M-I/R injury. lncRNA FGD5-AS1 was involved in the development of human cancers. For instance, silencing of lncRNA FGD5-AS1 inhibited the progression of non-small cell lung cancer by regulating the miR-493-5p/DDX5 axis (7). lncRNA FGD5-AS1 promoted tumor growth by regulating MCL1 via sponging miR-153-3p in oral cancer (8). However, the function and regulatory mechanism of lncRNA FGD5-AS1 remains unclear in the pathogenesis of M-I/R injury. lncRNAs have “sponge-like effects” on numerous miRNAs, including lncRNA FGD5-AS1. In this study, miR-129-5p was found to have a binding site with lncRNA FGD5-AS1. Overexpression of miR-129-5p mitigate sepsis-induced acute lung injury by targeting High Mobility Group Box 1 (9). More importantly, miR-129-5p protects H9c2 cardiac myoblasts from hypoxia/reoxygenation injury by targeting TRPM7 and inhibiting NLRP3 inflammasome activation (10). In addition, lncRNA NEAT1 promoted myocardiocyte apoptosis and suppressed proliferation through regulation of miR-129-5p (11). However, little is known about the interaction between lncRNA FGD5-AS1 and miR-129-5p in myocardial I/R injury. In the present study, the expression level of lncRNA FGD5-AS1 was detected in mouse I/R models. Then, the biological effect of lncRNA FGD5-AS1 in H9c2 cell H/R model was investigated. In addition, the regulatory mechanism of lncRNA FGD5-AS1/miR-129-5p in M-I/R injury was discovered.
Male c57BL6/J mice (20–24 g, Guangdong Medical Laboratory Animal Center) were feed in a standard pathogen-free environment (25°C, 60% humidity). Food and water were freely provided. All procedures were performed in accordance with the Care and Use of Laboratory Animals issued by the Chinese Association for Laboratory Animal Care and approved by our Hospital.
The myocardial I/R injury model was established by ligating the left anterior descending coronary artery. C57BL/6 mice were anesthetized with 3% pentobarbital sodium and a longitudinal incision. The mouse's thoracic cavity was open by left thoracotomy. Ligation was performed at approximately 3 mm from the source of the descending left anterior coronary artery with line 6–0. After 30 min induction of ischemia, the ligature was untied. Then, the mice were reperfused at various time points (6h, 12h, 24h).
The cardiomyocytes cell line H9c2 (Chinese Academy of Sciences, China) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) with 10% fetal bovine serum (FBS) in a humid incubator with 5% CO2 at 37 °C. H/R treatment was used to establish an I/R injury model in H9c2 cells. H9c2 cells were exposed for 24 h under hypoxia (5% CO2, 95% N2) and then re-oxygenated (5% CO2, 95% O2) for 12 h at 37°C. Control cells were incubated under normoxic conditions (NC).
FGD5-AS1 siRNA and vector or miR-129-5p mimics and inhibitor were designed and synthesized by GenePharma (Shanghai, China). They were transfected into H9c2 cells using Lipofectamine® 2000 transfection reagent, respectively.
When the M-I/R injury mice model was established, the levels of CK-MB and LDH were measured by using commercial assay kits (Invitrogen, Carlsbad, CA) in accordance with the manufacturer’s protocol.
Total RNA was extracted using TRIzol reagent (Thermo Fisher Scientific, MA, USA). Then, RT was conducted using a RevertAid First Strand cDNA Synthesis kit (K1622; Thermo Fermentas, USA). PCR was performed on an ABI Prism 7900 detection system (Thermo Fisher Scientific, Inc.) using iQ™ SYBR®-Green SuperMix (Bio-Rad Laboratories, Inc., Hercules, CA). GAPDH was applied as internal reference. MiRNA and mRNA expression levels were quantified using the 2−ΔΔCq method.
H9c2 cells (2×103cells/well) were seeded into 96-well plates and cultured for 12 h in 5% CO2 at 37°C. Then, cells were subjected to H/R exposure. Next, the cells were incubated with 15 μL/well MTT solution (5 mg/mL, Sigma) at 37 °C for 4 h. The absorbance value was determined at a wavelength of 490 nm by using a Bio-Rad 680 microplate reader (Bio-Rad Laboratories, Inc.).
H9c2 cells were collected and lysed by RIPA Lysis Buffer (Beyotime, Shanghai, China). Protein concentration was measured using Enhanced BCA Protein Assay kit (Beyotime, Shanghai, China). Next, protein samples (40 μg) were separated by 10% SDS-PAGE and transferred to PVDF membranes. The membranes were blocked with 5% skimmed milk for 2 h at room temperature and incubated with Bax, Bcl-2 and GAPDH primary antibodies (Abcam, Shanghai, China) overnight at 4°C. After washing, protein samples were incubated with horseradish peroxidase-conjugated secondary antibodies (Abcam, USA) for 2 h. Finally, the blots were detected using an enhanced chemiluminescence (ECL) reagent and analyzed with ImageJ software.
Wild-type and mutant FGD5-AS1 containing miR-129-5p binding site were amplified and inserted into the pGL3 vector (Promega) to construct recombinant reporter plasmids WTFGD5-AS1 and MUT-FGD5-AS1. The reporter plasmids and miR-129-5p mimics or miR-NC was co-transfected into H9c2 cells. After 48 h, dual luciferase assay system (Promega, USA) was used to detect luciferase activities.
Data were analyzed SPSS 19.0 (IBM Corp., Armonk, NY, USA) and expressed as mean ± SD. Graphs are made by Graphpad Prism 6. Student t-test was adopted to compare the difference between two groups, and multiple comparison was performed by one-way analysis of variance followed by Tukey’s post hoc test. P < 0.05 indicates statistically significant difference.
To explore the expression level of lncRNA FGD5-AS1 in myocardial I/R injury, myocardial I/R injury mice models were established. To assess whether the myocardial I/R injury mice mouse model is successfully established, the serum levels of LDH and CK-MB were detected. LDH and CK-MB serum levels were significantly increased in myocardial I/R injury group (Fig. 1A, 1B). Especially, the highest serum levels of LDH and CK-MB were occurred at 12 h after reperfusion (Fig. 1A, 1B). The myocardial I/R injury mice mouse model was successfully established. Next, RT-qPCR showed that lncRNA FGD5-AS1 expression was apparently increased at 6h and 12 h after reperfusion in myocardial I/R injury group compared with Normoxia group (Fig. 1C).
To explore the role of lncRNA FGD5-AS1 in myocardial I/R injury, H9c2 cell H/R model was established. RT-qPCR showed that lncRNA FGD5-AS1 expression was upregulated in H9c2 cells treated with H/R compared with Normoxia group (Fig. 2A). After transfection of FGD5-AS1 siRNA, FGD5-AS1 expression was reduced in H/R treated H9c2 cells, but still higher than that in Normoxia group (Fig. 2A). Functionally, cell proliferation was suppressed in H9c2 cells treated with H/R compared to Normoxia group. Downregulation of FGD5-AS1 promoted cell proliferation in H/R treated H9c2 cells. However, H9c2 cell proliferation in H/R + FGD5-AS1 siRNA group was still inhibited compared to Normoxia group (Fig. 2B). Additionally, the effect of lncRNA FGD5-AS1 on apoptosis-related protein (Bcl-2/Bax) was also detected in H9c2 cells. Compared with the Normoxia group, increased expression of Bax and decreased expression of Bcl-2 were identified in H/R group. Compared with H/R group, knockdown of FGD5-AS1 reduced Bax expression and enhanced Bcl-2 expression. The expression of Bax and Bcl-2 in H/R + FGD5-AS1 siRNA group tended to the levels in Normoxia group but could not reach the levels in Normoxia group (Fig. 2C, 2D). Briefly, lncRNA FGD5-AS1 could aggravate myocardial I/R injury by suppressing cell viability and inducing apoptosis.
To explain the regulatory mechanism of lncRNA FGD5-AS1 in myocardial I/R injury, the target of lncRNA FGD5-AS1 was searched in the star-Base database (http://starbase.sysu.edu.cn). We found that lncRNA FGD5-AS1 has a binding site with miR-129-5p (Fig. 3A). Dual-luciferase reporter assay showed that miR-129-5p mimics reduced the luciferase activity of wt-FGD5-AS1 (Fig. 3B), indicating that miR-129-5p is a direct target of lncRNA FGD5-AS1. Next, RT-qPCR showed thta miR-129-5p expression was reduced by FGD5-AS1 vector and promoted by FGD5-AS1 siRNA in H9c2 cells (Fig. 3C). At the same time, lncRNA FGD5-AS1 was upregulated in H9c2 cells with miR-129-5p mimics and down-regulated in H9c2 cells with miR-129-5p inhibitor (Fig. 3D).
To investigate the interaction between lncRNA FGD5-AS1 and miR-129-5p in myocardial I/R injury, miR-129-5p mimics and miR-129-5p mimics+FGD5-AS1 vector were transfected into H/R treated H9c2 cells. Compared with the Normoxia group, miR-129-5p was downregulated in H/R group. Upregulation of FGD5-AS1 reduced the increased expression of miR-129-5p induced by miR-129-5p mimics. However, miR-129-5p expression was still lower than that in Normoxia group (Fig. 4A). MTT assay showed that miR-129-5p overexpression promoted H/R treated H9c2 cell viability compared with H/R group. However, FGD5-AS1 vector weakened the promoting effect of miR-129-5p overexpression on H/R treated H9c2 cell viability (Fig. 4B). Compared to H/R group, overexpression of miR-129-5p reduced Bax expression and promoted Bcl-2 expression in H/R treated H9c2 cells. However, upregulation of FGD5-AS1 increased Bax expression and reduced Bcl-2 expression in H/R treated H9c2 cells compared with H/R+ miR-129-5p mimics group (Fig. 4C, 4D). These results indicate that overexpression of miR-129-5p ameliorates myocardial I/R injury. lncRNA FGD5-AS1 aggravates myocardial I/R injury by downregulating miR-129-5p.
Recently, many lncRNAs and microRNAs have been reported to participate in the pathogenesis of myocardial injury. For example, Downregulation of lncRNA NEAT1 promoted cell proliferation and inhibited cell apoptosis by targeting miR-193a in myocardial I/R injury (12). In this study, lncRNA FGD5-AS1 was upregulated in myocardial I/R injury mice models and H/R treated H9c2 cells. Functionally, knockdown of lncRNA FGD5-AS1 promoted cell viability and inhibited apoptosis in H/R treated H9c2 cells. In addition, miR-129-5p was confirmed to be a target of lncRNA FGD5-AS1. The protective effect of miR-129-5p on myocardial I/R injury was impaired by upregulation of lncRNA FGD5-AS1. These results demonstrate that lncRNA FGD5-AS1 aggravates myocardial I/R injury by down-regulating miR-129-5p. Consistent with our results, other lncRNAs also have been found to regulate myocardial I/R injury. For example, lncRNA FOXD3-AS1 aggravated I/R injury of cardiomyocytes through promoting autophagy (13). However, the role of lncRNA FGD5-AS1 has not been reported in myocardial I/R injury. Most studies reported that lncRNA FGD5-AS1 play important roles in human cancers. For instance, upregulation and carcinogenesis of lncRNA FGD5-AS1 has been detected in colorectal cancer and glioblastoma (14,15). However, lncRNA FGD5-AS1 expression was found to be decreased in oxygen-glucose deprivation and simulated reperfusion (OGD/R)-induced neurons injury. Up-regulation of FGD5-AS1 could recover proliferation and inhibit apoptosis of OGD/R-injured neurons (16). These results are contrary to our results in this study. This difference may be caused by different experimental materials. In the present study, lncRNA FGD5-AS1 directly targeted miR-129-5p and had a negative correlation with miR-129-5p expression in cardiomyocytes. More importantly, overexpression of miR-129-5p ameliorated myocardial I/R injury. Consistent with our study, miR-129-5p alleviates myocardial injury after ischemia/reperfusion (17). miR-129-5p ameliorated ischemia-reperfusion injury by targeting HMGB1 in myocardium (18). All these results indicate that miR-129-5p play a positive effect on myocardial I/R injury. In addition, upregulation of lncRNA FGD5-AS1 impaired the protective effect of miR-129-5p on myocardial I/R injury. lncRNA FGD5-AS1 could aggravate myocardial I/R injury. Interaction between lncRNA FGD5-AS1 and miR-129-5p in myocardial I/R injury has not been found in previous studies.
Upregulation of lncRNA FGD5-AS1 is detected in myocardial I/R tissues and cardiomyocytes. Upregulation of lncRNA FGD5-AS1 reduced cell proliferation and induced apoptosis in H/R treated cardiomyocytes. More importantly, lncRNA FGD5-AS1 aggravates myocardial I/R injury by downregulating miR-129-5p. Our results may provide a novel therapeutic or diagnostic target for myocardial I/R injury.
Ethical issues (Including plagiarism, informed consent, misconduct, data fabrication and/or falsification, double publication and/or submission, redundancy, etc.) have been completely observed by the authors. |
|
PMC9647626 | Kusuma Sai Davuluri,Devendra S. Chauhan | microRNAs associated with the pathogenesis and their role in regulating various signaling pathways during Mycobacterium tuberculosis infection | 27-10-2022 | miRNA,tuberculosis,pathogenesis,signaling pathways,anti-TB treatment | Despite more than a decade of active study, tuberculosis (TB) remains a serious health concern across the world, and it is still the biggest cause of mortality in the human population. Pathogenic bacteria recognize host-induced responses and adapt to those hostile circumstances. This high level of adaptability necessitates a strong regulation of bacterial metabolic characteristics. Furthermore, the immune reponse of the host virulence factors such as host invasion, colonization, and survival must be properly coordinated by the pathogen. This can only be accomplished by close synchronization of gene expression. Understanding the molecular characteristics of mycobacterial pathogenesis in order to discover therapies that prevent or resolve illness relies on the bacterial capacity to adjust its metabolism and replication in response to various environmental cues as necessary. An extensive literature details the transcriptional alterations of host in response to in vitro environmental stressors, macrophage infection, and human illness. Various studies have recently revealed the finding of several microRNAs (miRNAs) that are believed to play an important role in the regulatory networks responsible for adaptability and virulence in Mycobacterium tuberculosis. We highlighted the growing data on the existence and quantity of several forms of miRNAs in the pathogenesis of M. tuberculosis, considered their possible relevance to disease etiology, and discussed how the miRNA-based signaling pathways regulate bacterial virulence factors. | microRNAs associated with the pathogenesis and their role in regulating various signaling pathways during Mycobacterium tuberculosis infection
Despite more than a decade of active study, tuberculosis (TB) remains a serious health concern across the world, and it is still the biggest cause of mortality in the human population. Pathogenic bacteria recognize host-induced responses and adapt to those hostile circumstances. This high level of adaptability necessitates a strong regulation of bacterial metabolic characteristics. Furthermore, the immune reponse of the host virulence factors such as host invasion, colonization, and survival must be properly coordinated by the pathogen. This can only be accomplished by close synchronization of gene expression. Understanding the molecular characteristics of mycobacterial pathogenesis in order to discover therapies that prevent or resolve illness relies on the bacterial capacity to adjust its metabolism and replication in response to various environmental cues as necessary. An extensive literature details the transcriptional alterations of host in response to in vitro environmental stressors, macrophage infection, and human illness. Various studies have recently revealed the finding of several microRNAs (miRNAs) that are believed to play an important role in the regulatory networks responsible for adaptability and virulence in Mycobacterium tuberculosis. We highlighted the growing data on the existence and quantity of several forms of miRNAs in the pathogenesis of M. tuberculosis, considered their possible relevance to disease etiology, and discussed how the miRNA-based signaling pathways regulate bacterial virulence factors.
Non-coding RNAs are single-stranded transcripts that regulate the mRNA (coding gene) expression by degrading them. microRNAs (miRNAs) are small molecules of non-coding RNA that contain 17–25 nucleotides and modulate gene expression (Esteller, 2011). From the past few decades, there is great progress in miRNA research, which is believed to be important in regulating various pathological processes (Aguilar et al., 2019). Over 20 years ago, the first miRNA discovery made a mark in a molecular biology new era. Over 2,000 miRNAs have been identified in humans, and it is thought that they collectively regulate one-third of the genes in the genome. miRNAs have been linked to a variety of human diseases and are being researched for use as clinical diagnostic and therapeutic targets through biogenesis, involving the multitude of mechanisms that inactive miRNA converts into mature miRNA (Taneja and Dutta, 2019). Disruption in the maturation process of miRNA, for example, miR-146a irregular expression and impaired miRNA regulatory mechanisms, leads to neoplasia, ischemic heart disease, neurodegenerative diseases, etc. Yang et al., 2019 (Chen et al., 2019). miRNA formation process occurs in the nucleus by RNA polymerase II. Initially, the primary transcript with hair pin structure that encodes miRNA sequences is regulated by RNA polymerase II transcription factors, epigenetic and histone modifiers. Primary miRNA goes through maturation processes by cropping the loop end of pri-miRNA (pre-miRNA) in the nucleus. Later, the resulting product is exported to the cytoplasm by exportin-5 for further maturation steps. RNase, Dicer crops the loop end one more time resulting in the small RNA duplex (Bogerd et al., 2014). Only six nucleotides that match are required to obtain functional miRNA (Bartel, 2009). Recent research studies reveal the genesis and role of miRNA in regulating several bacterial pathogenesis-associated signaling pathways (Zhang et al., 2017b; Stutz et al., 2018). miRNAs can be reliable in therapeutic settings. These factors became important in screening the diseases with high specificity, sensitivity, and accessibility (Walker and Harland., 2009 Ostrik et al., 2021). Northern blotting, microarray analysis, and quantitative polymerase chain reaction (qPCR) are traditional methods for miRNA detection. To improve the sensitivity and selectivity of miRNA detection, new technology methods always rely on signal amplification strategies, such as nanoparticle-based amplification, isothermal exponential amplification, rolling circle amplification, hybridization chain reaction, and combinations of these (Ye et al., 2019). Our literature review reveals the role of miRNAs as modulators of signaling pathways in tuberculosis (TB). miRNAs act as genetic switches that make them regulators of cellular signaling pathways. We can predict the targets of miRNA easily nowadays through the discovery of high-throughput genomic screening methods. Understanding the role of miRNA in signaling pathways might lead to novel therapies. New kinase inhibitors are being studied to treat many diseases by detailed understanding of the role of miRNA in regulating the kinase cascade pathways. We will indeed be able to create new therapeutical platforms, such as locked genomic technology, for synthesizing and providing efficient RNA-based chemotherapeutic agents. miRNA expression patterns differ in active TB, latent tuberculosis infection (LTBI), and healthy individuals (Sabir et al., 2018). miRNA synthesis mainly influences the action of various immune cells (Chandan et al., 2020). We summarize some of them and discuss their benefits and drawbacks for improving miRNA detection design. Research studies found significant variations in miRNA patterns that help in identifying LTBI and long-term TB infection. Compared to LTBI, active tuberculosis shows upregulation of miR-194-5p, miR-21, miR-29c-3, miR-150-5p, miR-365a-3p, miR-223-3p, miR-451a-5p, miR-44-5p, and miR-144-3p (Wang et al., 2011). Innate immune response: miR-146a (IRAK)-1/ (TRAF)-6], miR-9 (NF-κB1), miR-125b (ERK)1 (Zhou et al., 2010), miR-26-5p (KLF4), miR-132-3p [(TLR)]; Regulation of inflammation: miR-21-5p (TLR4), miR-146a-5a (TRAF-6), miR-20b-5p (NLRP3), miR-223-3p (NFIA), miR-27b-3p (Bag2), miR-99b-5p [(TNF)-α and TNF receptor superfamily (TNFRSF)-4], miR-125-5p (TNF-α), miR-142-3pN (Wasp), miR-144(IFN-γ and TNF-α), miR-27a (IRAK-4); Autophagy: miR-155 (Rheb), miR-27a (Cacna2d3), miR-889 (TWEAK), miR-106a (ULK1, ATG7, ATG16L1) (Yang J et al., 2019), miR-125 (DRAM2), miR-142-3p (ATG16L1) (Yang Y et al., 2019), miR-17 (ATG7), miR-144-3p (ATG4a), miR-20a (ATG7/ATG16L1) Cui et al., 2022a, miR-23a-5p (TLR2/MyD88/NF-κB), miR-26a (KLF4); Apoptosis: miR-27a, miR-96 (FOXO3) (Guttilla and White, 2009), miR-20a-5p (JNK)2, miR-27b (Bag2), miR-21 [PI3K/Akt NF-κB], Let-7e (Caspase-3), miR-29a (Caspase-7) (Pattnaik et al., 2022). In this review, we focused on the six main signaling pathways involved in the major pathogenic mechanisms such as autophagy, inflammation, and apoptosis. miRNAs that show strong research evidence of regulation according to the target scan and miRbase software were discussed. Finding out the function of various miRNAs in the regulation of various pathogenic signaling pathways may lead to identifying new therapeutic targets. Inactive mRNA undergoes splicing/processing to convert into mature mRNA. Mature mRNA then transported from nucleus to cytosol where it is translated as shown in Figure 1 .
miRNA expression patterns in patients with active TB were shown to be distinct from those of individuals with LTBI or healthy controls (Fu et al., 2011; Harapan et al., 2013; De Araujo et al., 2019). miRNA synthesis may influence the activation of natural killer cells, macrophages, dendritic cells, and T cells (Xu et al., 2019). To avoid the immune system, Mycobacterium tuberculosis may either enhance or inhibit miRNA expression. TNF-α and interferon (IFN)-γ are the host cytokines associated with autophagy during bacterial infection. Myeloid cells triggered by TLR signaling have been demonstrated to be negatively affected by higher levels of miRNA-146a-5p, miR-21-5p, miR-155, miR-199b, and miR132-5p (Kim et al., 2017). The overexpression of miR-27a-5p and miR-33 in M. tuberculosis-infected cells inhibits the creation of autophagosomes and the killing of M. tuberculosis by macrophages (Liu et al., 2018). M. tuberculosis-infected macrophages overexpress miRNAs that target IFN-γ and TNF-α, which suppress the immunological response against M. tuberculosis (Chakrabarty et al., 2019). As an additional line of defense against intracellular infections, host miRNAs such as miR-325-3p and miR-20b-5p impact cell death and inflammasome activation (Lou et al., 2017; Fu et al., 2020). Host innate and adaptive immune systems, such as miR-155-5p and let-7f, both have a role in the activation of miRNAs during M. tuberculosis infection, which is necessary for the clearance of pathogens (Rothchild et al., 2016). In cohorts that comprised persons with LTBI, active TB, and healthy controls, researchers have studied miRNA expression profiles in serum/plasma or blood cells (Lyu et al., 2019). Compared to LTBI, the expression of five miRNAs was higher in TB patient PBMCs (miR-365a-3p, miR-223-3p, miR-451a-5p, miR-44-5p, and miR-144-3p), with target predictions pointing to a possible role in TB patients’ hematopoiesis. According to another research, active TB patients had an improved expression of miR-194-5p and other miRNAs, including miR-21, miR-29c-3, and miR-150-5p. Upregulation of miR-29a-3p was shown to be a helpful prospective biomarker for qRT-PCR-based differentiation among active TB and LTBI (Zhang et al., 2014; Kanniappan et al., 2017). Individuals infected with HIV and those who were not exhibited similar levels of miRNAs. It was observed that miR-1246, miR-2110, miR370-3p, miR-28-3p, and miR-193b-5p were overexpressed in active TB, whereas miR-3675-5p was downregulated (Duffy et al., 2018). There was no validity testing done on the patients in the second cohort. Pathological Biomarkers for Tuberculosis Progression and Therapy Response researchers want to find predictive miRNA signatures for LTBI-to-TB progression and anti-TB medication response. According on published data rather than a screening in the lab, these miRNAs were selected. TB patients who received successful TB treatment were shown to have lower levels of other miRNAs than those who did not react to treatment (Lyu et al., 2019). While the concept and therapeutic regimen were the same, the screening method was different in a Chinese study. Compared to miR-148b-3p, miR-92a-3p, and miR-21-5p, miR-125a-5p was elevated in this case (Zhao et al., 2013; Duffy et al., 2018). Due to discrepancies in results, standardization of screening techniques is needed to provide more accurate results. In patients with active TB, LTBI, and isoniazide-treated LTBI, researchers found three miRNAs (let-7a-5p, a small nucleolar RNA miR-196b-5p, and SNORD104) as highly sensitive classifiers to distinguish TB from non-TB group members using insignificant RNA sequencing (RNA-seq) of whole blood (Barry et al., 2018). Regardless of the prevalence of HIV-1 coinfection, small RNA levels in plasma dropped dramatically before and after therapy. Although miR-29a-3p, SNORD61, miR-17-3p, and miR-133a levels were reduced among persons who reacted to medicine compared to those who did not, no single miRNA or combination of small RNAs was shown to be a significant predictor of successful TB therapy (Wang et al., 2018). To investigate whether there was a similar profile of differential miRNA expression across trials from patients with active TB and healthy controls, samples from patients with active TB were compared to those of healthy controls. Since the past decade or so, researchers have used this method to find miRNA markers in serum/plasma and blood cells. miRNAs described employing broad-spectrum unbiased procedures such as small RNA-seq will likely provide new accurate results than researchers who just concentrate on a few possible miRNAs. Because only a few miRNAs are available in the signature revealed by two or more studies, there is a lack of consistency in the outcomes of such screenings. An array of patient demographics and different types of RNA-seq and microarrays may be at fault. It is thought that these miRNAs play an essential role in TB pathogenesis by decreasing the host’s innate and acquired immune response to intracellular infections, both directly and indirectly. Anti-inflammatory miRNAs, such as miR-21-5p and miR-146a-5p, may also be used to discriminate among active TB and LTBI or an otherwise wholesome condition (Spinelli et al., 2013). As a putative biomarker of active TB, the M. tuberculosis inducing miR-155-5p, which is overexpressed in sufferers, plays a vital part in host defense (Etna et al., 2018). Role of different miRNAs in the pathogenesis of tuberculosis is tabulated in Table 1 .
AMP-activated protein kinase (AMPK) is a crucial metabolic sensor that responds to alternations in the cellular AMP/ATP ratio following the activation of catabolic energy production. The AMPK pathway is also activated in response to various bacterial infections and inflammation. Various bacterial antigens activate AMPK signaling cascades associated with host response modulation that can either increase or decrease pathogen survival (Grahame Hardie, 2016; Prantner et al., 2017). Previous research has revealed a wide range of AMPK pathway functions, including the regulation of host signaling and participation in significant events. mTOR kinase phosphorylation was more activated in macrophages than AMPK in a time-dependent manner following M. tuberculosis infection (Yang et al., 2014). There is evidence that cytosolic M. tuberculosis colocalizes with p62 and LC3, which are autophagic machinery components. Watson et al. 2012 found it in only 30% of total M. tuberculosis phagosomes, providing evidence that the majority of intracellular M. tuberculosis could inhibit xenophagy activation because of TFEB nuclear translocation downregulation. Activation of mTOR during. M. tuberculosis infection raises the levels of miRNA-33 and miRNA-33a (Oneyama et al., 2011). miR-124 reduces cell proliferation through G1 phase cell cycle arrest (Gong et al., 2016; He et al., 2020). Furthermore, overexpression of miR-124 overexpression reduces both cell growth and glucose consumption in cells (Zhao et al., 2017), which is consistent with AMPK downregulation. Increased levels of miR-101a and miR-199a decrease AMPK signaling (Liu et al., 2016b) (Li et al., 2020), and miR-101a can both directly and indirectly inactivate AMPK (Li et al., 2018; Liu et al., 2018). miRNAs that were downregulated during hypoxia are also involved in AMPK signaling, which plays an important role in reducing oxidative stress, autophagy, and apoptosis during hypoxia (Li et al., 2016; Tran et al., 2017; da Cruz et al., 2018; Sun et al., 2018; Zhu et al., 2018; Zhao T. et al., 2020). Starvation, genotoxic stress, hypoxia, ER stress, and reactive oxygen species (ROS) all activate signaling pathways that either initiate or regulate autophagy cascades. AMPK-mTORC1 regulates autophagy by integrating multiple stimuli and pathways into a signal for the ULK complex, which is the starting point for autophagy. Several miRNAs have been identified as regulators of AMPK-mTORC1. miRNAs acts as both positive and negative regulators of the gene expression. Upregulated miRNAs mainly target the signaling pathways associated with pathogenesis during the infection. The miRNAs targeting the various pathways are shown in the Figure 2 .
The host component against pathogenic organisms is NF-κB, which is a regulator of cell pro-inflammatory responses (Zhang Q et al., 2017a). NF-κB has been linked to the emergence of chronic inflammation and bacterial infections. Recent studies have reported that several miRNAs have been associated with the inflammatory responses by regulating the M. tuberculosis replication and induced pathogenesis by targeting the TRAF-6 signaling pathways (Cui et al., 2018). The detailed mechanisms have to be illustrated further. TRAF-6 belongs to the TNF receptor protein family, acts as an important regulator in many cellular pathways and regulates signal transduction of the TNF receptor superfamily (Chen et al., 2020). TRAF-6 also acts as a link among IRAK-1/IRAK and NF-κB/IB kinase signaling pathways in response to pro-inflammatory cytokines by binding with TGF-β-activated kinase-1 (TAK1) and supporting IκB kinase phosphorylation, ubiquitination, and deterioration after the stimulation of many innate immunity-associated genes (Li H. et al., 2018). During the M. tuberculosis infection, upregulation of miRNA-125a in macrophages depends on TLR4 signaling by targeting TRAF-6 and modulating NF-κB (Cui J et al., 2022a; 2022b). This mechanism leads to the attenuation of the immune response and enhances the survival of mycobacteria. The NF-κB pathway is associated with bacteria–host interactions (Westermann, 2018). By identifying the PI3K-AKT-mTOR signaling pathway (PTEN), miR-26b tends to promote the LPS-induced NF-κB signaling pathway and enhances the expression of pro-inflammatory factors (Huang et al., 2012). Studies showed that genetic disruption of the p50 subunit of NF-κB restricts the M. tuberculosis infection. Pharmacologic regulation of NF-κB activation decreases the viability of intracellular mycobacteria (Liu et al., 2016a). NF-κB inhibition increases the apoptosis of macrophages and autophagy, which is the established defense mechanism (Bai et al., 2013). Inactivation of NF-κB downregulates the expression of PTEN that regulates cellular activities that may be crucial for pathogen resistance. PTEN signaling regulates infection by affecting various intracellular mycobacterial pathogens (Fang et al., 2016). PTEN deficiency renders susceptibility to infection in multiple cells infected with mycoplasma and Mycobacterium. PTEN’s lipid phosphatase activity is critical for infection tolerance. Mycobacterium infectious disease activates Akt phosphorylation, and suppression of Akt or PI3K activity regulates cellular infection (Huang et al., 2012). M. tuberculosis-infected macrophages secrete cytokines, showing an effective defense mechanism against the pathogen. NF-κB and mitogen-activated protein kinase (MAPK) signaling pathways regulate the expression of various cytokines (Gañán-Gómez et al., 2014). Cytokines such as TNF-α, IL-6, and IL-1β are potent mediators showing immune response against the M. tuberculosis bacilli (Cui et al., 2021). Targeting the cytokines and their regulatory pathways restricts the host immune response. For example, M. tuberculosis virulence protein PtpA arrests the NF-κB and JNK signaling pathways (Rothchild et al., 2016). Treatment of cells with early secreted antigenic target-6 (ESAT-6) prevents TLR-associated NF-κB activation (Yang et al., 2015). TAK is a serine/threonine protein kinase associated with the activation of NF-κB pathway. Studies show that miRNAs regulate TAK expression to promote chemoresistance. Upregulation of miR-143 attenuates the function of TAK. miR-146a and miR-26b also target the TAK to promote apoptosis and are associated with the NF-κB pathway inhibition. miR-143 and miR-146a inhibit the NF-κB signaling pathway (Chen Y et al., 2016; Chen et al., 2019).
During the M. tuberculosis infection, TLR2-deficient animals were more susceptible than control mice, but TLR2- and TLR4-deficient mice were as vulnerable as control mice in a low-dose M. tuberculosis challenge (Ju et al., 2018). Pattern recognition receptors (PRRs) expressed on leukocytes activate phagocytosis and host defense mechanisms through the promotion of signaling cascades. TLRs and mannose receptors associated with PRRs play a critical role in immune response and detect the pathogen-derived molecules. Most of the mycobacterial antigens act as agonists for TLRs (Liu Y. et al., 2016). Inoculation of BCG was believed to be dependent on TLR2 and TLR4. Most of the mycobacterial proteins and lipids are associated with the TLR-dependent signaling cascades. TLR regulates hundreds of the host genes that are associated with signaling and acts against microbial antigens, so studying about TLR molecular mechanisms has great importance (Shariq et al., 2021). TLR signaling was negatively affected by higher levels of miRNA-146a-5p, miR-21-5p, miR-99b-5p, and miR132-5p 9 (Wu et al., 2012). The overexpression of miR-27a-5p and miR-33 in M. tuberculosis-infected cells inhibits the creation of autophagosomes and the killing of M. tuberculosis by macrophages (Marcinowski et al., 2012). As an additional line of defense against intracellular infections, host miRNAs such as miR-325-3p and miR-20b-5p impact cell death and inflammasome activation (Kumar et al., 2015). miRNAs associated with the host innate and adaptive immune systems, such as miR-155-5p and let-7f, have a role in the activation of signaling pathways during M. tuberculosis infection, which is necessary for the clearance of pathogens (Iwai et al., 2015; Li et al., 2016). Although miR-29a-3p, SNORD61, miR-17-3p, and miR-133a levels were reduced among people who reacted to medicine compared to those who did not, no single miRNA or combination of small RNAs was shown to be a significant predictor of successful TB therapy (Dersch et al., 2017). M. tuberculosis-mediated TLR2/1 signaling increases the expression of the vitamin D receptor and the vitamin D hydroxylase, resulting in enhanced production of antimicrobial peptides (Lv et al., 2017). TLR4 may contribute to M. tuberculosis resistance; however, no agreement has been achieved at this point. TLR4 has a protective role in adaptive immunity against pulmonary TB in vivo; the non-functional TLR4 causes high mortality and increased bacterial burden in the lungs. miR-146a-5p, miR-21-5p, miR-99b-5p, and miR-132-5p are highly expressed in TB patients and adversely regulate host signaling cytokines in myeloid cells triggered by TLR signaling, promoting M. tuberculosis survival (He et al., 2018). Other miRNAs that are upregulated in M. tuberculosis-infected macrophages, such as miR-27a-5p, miR-33, miR-125-5p, and miR-144-5p, inhibit autophagy formation and M. tuberculosis killing by macrophages. Both miR-29a-3p and miR-125-5p are upregulated in infected macrophages and directly target IFN and TNF, thereby reducing the immune reaction to intracellular M. tuberculosis (Stepanov et al., 2015). Cell necrosis and inflammasome formation are two other mechanisms of defensive strategy against intracellular pathogens that are controlled by M. tuberculosis-induced host miRNAs such as miR-325-3p and miR-20b-5p. However, some miRNAs that are influenced during M. tuberculosis infection, such as miR-155-5p and let-7f, play a crucial role in the activation of host innate and adaptive immunity, as well as microbial clearance (Sinigaglia et al., 2020).
Ligand identification activates the TIR-containing adaptor receptor MyD88, which further binds to IRAK-1 and IRAK-4 (Li et al., 2013; Gu et al., 2017). IRAK has a destruction domain and a serine/threonine kinase domain, and there are four members of the IRAK family: IRAK-1, IRAK-2, IRAK-M, and IRAK-4 (Li et al., 2002). Studies have revealed that IRAK-4 functions upstream of IRAK-1 in the TLR complex (Zhang et al., 2019). Mutations in the IRAK-4 gene have been linked to a higher sensitivity to bacterial infection in patients with Mendelian susceptibility to mycobacterial disease, and M. tuberculosis-infected patients are resistant to TLR ligands (Cui et al., 2018). Moreover, NF-κB essential modulator (NEMO) and IRAK-4 were revealed to be important in IL-12 formation and increased IFN-γ production in humans and mice, which would be vital to creating protective immune responses against mycobacterial infection (Wu et al., 2019). In response to TLR stimulation, IRAK-4 associates with IRAK-1, and the emergence of a dominant negative form of IRAK-4 negative regulator IRAK-1 activation (Li S et al., 2002). IRAK-deficient mice secrete more cytokines in response to TLR ligands (Lomaga et al., 1999). IRAK-M is the negative regulator of the TLR signaling that shows the important role of this protein in suppressing mycobacteria-induced inflammasome activation and TLR signaling pathways. The MAPK pathways are triggered by primary stimulations such as mycobacterial products or whole mycobacteria, resulting in the stimulation of transcription factors such as NF-κB and activator protein-1 (AP-1) (Wang et al., 2001). miRNAs regulating the IRAK signaling pathway were shown in Table 2 .
Death ligands such as Fas ligand (FasL) bind to death receptors in the FasR receptor. Following this interaction, the death-inducing signaling complex (DISC) is formed, which includes the Fas-associated death domain-containing protein (FADD) and procaspase-8/10. RNAi-mediated FADD knockdown in cells reduced NF-κB signaling. Exogenous FADD expression prevented NF-κB signaling (Dockrell, 2003). FADD loss-of-function mutations in the death effector inhibited the caspase-8 and NF-κB activation that promotes apoptosis. Caspase-8 deficiency inhibited TNF-related apoptosis-inducing ligand (TRAIL)-induced NF-κB activation. These findings reveal a mechanism for TRAIL-induced NF-κB activation that involves the TRAIL receptors DD, FADD, and caspase-8. These proteins play an important role in apoptosis signaling and are the mediators of non-apoptotic CD95 signaling during T-cell proliferation (Welz et al., 2011). FADD-deficient T cells show reduced proliferation, implying that FADD plays an important role in proliferation signaling. FADD can be regulated transcriptionally by miR-155 (Wang et al., 2011) or miR-128a. miR-128a ectopic expression conferred Fas resistance in cells by directly targeting FADD, but antagonizing miR-128a function made cells susceptible to Fas-mediated apoptosis (Yamada et al., 2014).
Following the binding of activated calcium ions to cAMP, they activate small Ras-like GTPases like Ras-proximate-1 (Rap1), which is primarily involved in cell adhesion and junction formation during cell proliferation. cAMP is also known to increase ERK1/2 phosphorylation via ROS-dependent activation of Ras. Through the negative feedback regulation, miR-146 plays an important role in the control of TLRs and cytokine signaling (Brooks et al., 2014). miRNAs have the ability to regulate the levels of molecules by being involved in the negative feedback of PRR-induced signaling (Mao et al., 2005). miR-124 was discovered to be a negative regulator of inflammation by targeting several pathways, including signal transducer and activator of transcription (STAT) and TLRs (O’Shea et al., 2013). miR-124 inhibits intestinal inflammation by attenuating the production of IL-6 and TNF-α via targeting STAT3, a major factor in inflammatory response, and acetylcholinesterase, a negative regulator of the cholinergic anti-inflammatory signal. Sun et al. (2018) reported that miR-124 inhibits STAT3 to reduce IL-6 production and TNF-α-converting enzyme to inhibit TNF-α release in response to LPS. Lower levels of miR-124 and higher levels of STAT3 promote inflammation and disease pathogenesis. miR-124 expression is increased in pulmonary TB patients. miR-124 negatively regulates multiple TLR signaling components, including TLR6, MyD88, TNF-α, and TRAF6, implying an underlying negative feedback loop between miR-124 and TLR signaling to prevent excessive inflammation (Wang S. et al., 2018). In both calves and humans, a decrease in miR-124 expression contributes to high proliferation and pulmonary inflammation.
MyD88 is an intracellular molecule connected to IRAK and TLRs to transduce signals. MyD88 activates MAPK, PI3K, NF-κB, and IRAK following the initiation of the signal cascade. MyD88 deficiency impairs the macrophage response to bacterial antigens and makes the individual susceptible to infection. However, macrophages activate antibacterial immunity through MyD88-independent mechanisms (Cervantes, 2017). MyD88 deficiency improved resistance to polymicrobial sepsis, indicating that both MyD88-dependent and MyD88-independent antibacterial mechanisms exist. Many studies that showed the regulation of individual genes in macrophages by subcellular microbial products through the TLR/MyD88 signal transduction pathway have been conducted (Huang et al., 2019). However, there appears to be no study showing the role of MyD88 in macrophage activation showing antimicrobial activity. Three unexpected findings emerged in MyD88-deficient mice, implying that the current understanding of macrophage activation needs to be revised. Macrophages undergo active self-priming activation, which is dependent on MyD88. MyD88 is not involved in the IFN-γ signaling pathway; however, the expression of many genes in macrophages in response to IFN-γ is mostly dependent on MyD88. The majority of transcriptional responses of macrophages against M. tuberculosis do not require MyD88. This suggests that TLRs are not the primary receptors for recognizing M. tuberculosis or that TLR-dependent responses are mediated by MyD88-independent signaling pathways (O’Connell et al., 2010; Sharbati et al., 2011). miR-155 modulates the production of inflammatory mediators in response to microbial stimuli by negatively regulating the expression of an important TAK1- and TRAF6-binding protein 2 (TAB2) (Ceppi et al., 2009). Additionally, miR-146a inhibits TLR signaling, thereby inhibiting the production of inflammatory mediators (Taganov et al., 2006; Chen et al., 2007). When PAMPs are recognized, TLR signaling is activated, which leads to the transcriptional activation of genes encoding pro-inflammatory mediators following the activation of antigen-specific adaptive immune response via a MyD88-dependent or -independent pathway (Medzhitov et al., 1998). Various signalling pathways are regulated by more than one miRNAs. The miRNAs that regulate the various signalling pathway during the tuberculosis infection are shown in the Table 2 .
The Bcl-2 family members that promote and prevent apoptosis are controlled differently by M. tuberculosis. The prototypical antiapoptotic protein Bcl-2 has homologs in the Bcl-2 family (Klingler et al., 1997). The antiapoptotic family member bfl-1 is upregulated in macrophages during the Mycobacterium bovis BCG infection (Perskvist et al., 2002). An antiapoptotic gene called Bcl-xL was upregulated during M. tuberculosis infection after Bcl-2 was downregulated (Mogga et al., 2002). Infection with M. tuberculosis causes neutrophils to produce more of the proapoptotic family protein Bcl-2-associated X-protein (Bax) and less of the antiapoptotic family protein Bcl-xL (Harris and Thompson, 2000). Bcl-2 was upregulated and Bax was downregulated in animal models of TB. By reducing neutrophil levels and increasing B-cell levels, a number of miRNAs have been linked to the regulation of the apoptotic pathway (van Rensburg et al., 2018). miR-365, which is highly expressed in cells, directly targets the proapoptotic protein Bax, and these interactions are linked to drug resistance in pancreatic cancer cells. By inhibiting Bax expression, miR-125b conferred drug resistance in breast cancer cells (Zhou et al., 2010). By specifically targeting Bcl-xL and inducing apoptosis, miR-491 reduces the viability of cells. Treatment with miR-491 prevents tumor growth in naive mice in vivo (Nakano et al., 2010). Downregulation of miR-133a has been linked to tumor development and prognosis. Restoration of miR-133a inhibits cell division and triggers apoptosis. Bcl-xL regulation by miR-608 has also been demonstrated (Zhang et al., 2014). The expression of Bcl-2 was discovered to be inversely correlated with miR-15a and miR-16-1 (Cimmino et al., 2005). These two miRNAs directly inhibit Bcl-2 at the posttranscriptional level, according to a subsequent study, and also cause apoptosis (Zhang Y. et al., 2014). Bcl-2 protein signaling was elevated when miR-204 was downregulated. MiR-148a and miR-24-2c also directly inhibit Bcl-2 expression (Srivastava et al., 2011; Zhang et al., 2011). Apoptosis is regulated by the endogenous miR-23a/b and miR-27a/b inhibitors of apoptotic peptidase-activating factor (Apaf)-1 expression. It has been demonstrated that miR-133 and miR-24a directly repress caspase-9 to regulate cell fate (Xu et al., 2007; Walker and Harland, 2009; Ji et al., 2013; Chen et al., 2014).
Caspases play an important role in classical apoptosis (Kroemer and Martin., 2005). Caspase activation is not necessarily important in all types of apoptosis. Apoptosis can be triggered by the extrinsic pathway and the intrinsic pathway involving ligation of cell surface death receptors through respective ligands and regulating the Bcl-2 family of pro- and antiapoptotic proteins, respectively. Suppression of caspase activity is cytoprotective when cells are stimulated to undergo apoptosis via death receptor ligation (Maquarre et al., 2005). Caspases, on the other hand, are terminal effectors of the mitochondrial pathway, and this type of cell death is mostly caspase independent (Jäättelä, 2001; Hudson et al., 2013). Few studies suggest that apoptotic cell death can occur in the absence of caspases or in the presence of both caspase and non-caspase protease activity (Sacconi et al., 2012). TNF-α activates caspases and can initiate several cell death pathways, including lysosomal permeabilization mediated by cathepsin B release, which activates the mitochondrial apoptosis pathway (Guicciardi et al., 2000). Although these caspase-independent pathways have not been fully characterized, calpains and serine proteases have been implicated as cell death mediators (Zhang L. et al., 2012; Guo et al., 2013). Overexpression of miR-337-3p and miR-17-5p/miR-132-3p/-212-3p, respectively, can regulate executioner caspase-3 and caspase-7. Furthermore, miRNA overexpression, particularly miR-337-3p, reduces TRAIL cytotoxicity.
MAPK-regulated ERK, JNK, and p38 groups alter gene expression. MAPK signaling pathways are activated during mycobacterial infection and are related to mycobacterial pathogenesis (Song et al., 2003; Pasquinelli et al., 2013; Guo et al., 2019). The p38 MAPK pathway is associated with mycobacteria-induced IL-10, and other cytokines such as TNF-α/IL-4/IFN-γ are produced (Bachstetter and Van Eldik, 2010; de Souza et al., 2014). TNF-α expression in human macrophages is increased by ERK1/2 signaling (Surewicz et al., 2004). MAPK signaling pathways are involved in the regulation of antimycobacterial pathways such as phagosome acidification, apoptosis, and antigen presentation via MHC class II expression. Previous research indicates that the p38 MAPK pathway could serve as a means for mycobacteria to be suppressed. Suppression of p38 MAPK activity increases phagosome acidification. Inactivation of the p38 MAPK pathway causes an increase in phagosome acidification and a significant increase in monocytes’ ability to kill mycobacteria (Klug et al., 2011). Synthesis of TNF in human macrophages is inhibited by lipomannan from virulent M. tuberculosis, but not by avirulent Myocobacterium smegmatis. This variation in response is due to TB and lipomannan induces the stimulation causing TNF mRNA transcripts to destabilize following the reduced expression of TNF protein. Mycobacterium smegmatis Lipomannan increases MAPK-activated protein kinase 2 (MK2) phosphorylation, which is important for maintaining TNF mRNA stability by contributing miRNAs. miR-125b binds to the 3’ UTR region of TNF mRNA and destabilizes the transcript, whereas miR-155 increases TNF production by increasing TNF mRNA half-life and reducing the expression of SHIP1, which is the negative regulator of the PI3K/Akt pathway (Rajaram et al., 2011). Signaling via ERK1/2 and p38 inhibits a well-known mycobacterial TLR2 agonist. Thus, the p38 MAPK and ERK1/2 pathways regulate macrophage antimicrobial function and antigen presentation by infected macrophages, potentially contributing to host immune evasion (Liu P et al., 2016; Liu P et al., 2016; Hölscher et al., 2020). TNF mRNA is stabilized by activated MK2 (Campbell et al., 2014). Non-phosphorylated TTP Tristetraprolin binds to the Adenine/Guanine rich elements ARE region of target mRNAs and causes rapid degradation via a variety of mechanisms (Chen et al., 2013). During the Mycobacterium infection, TLR2-dependent MAPK p38 and the PI3K/Akt pathway stimulates an increase in TNF mRNA expression. Mycobacteria cause the activation and expression of MK2, miR125b, and miR-155 to differ. TNF expression in mycobacteria-infected macrophages is significantly influenced by MAPK p38 and Akt activation. miR-125b inhibits TNF production by targeting the 3′ UTR of the TNF transcript. It also increases the stability of B-Ras2, an inhibitor of NF-κB signaling in human macrophages, lowering the inflammatory response (Niu et al., 2018). TNF production is regulated by hsa-miR-155, which targets the inositol phosphatase for degradation via its 3′ UTR interaction (Yang et al., 2015). Mycobacteria cause differential expression of miRNAs, which are involved in mRNA signal transduction in human macrophages (Schifano et al., 2017). The JNK signaling pathway is important in many biological processes, including embryogenesis. These kinases regulate the expression of host genes involved with apoptotic cell death pathways and carcinogenesis, thereby controlling the functions of neurons and the immune system. Several miRNAs and long noncoding RNAs (lncRNAs) are functionally related to JNKs (Ghafouri-Fard et al., 2021). miR-138 targets mixed-lineage kinase-3 (MLK3), an important component of the JNK/mitogen-activated kinase pathway. miR-138 upregulation diminished proapoptosis factors and apoptosis rate. Upregulation of miR-138 decreased the expression of JNK, phosphorylated JNK (p-JNK), c-jun, p38 MAPK, p-p38 MAPK, iNOS, and COX-2 (Ghafouri-Fard et al., 2021). Low concentrations of MLK3 proteins and inhibition of the JNK/MAPK signaling pathways provide protection. miRNA-363-3p transcriptional regulation is mediated by DNA methylation. The dual-specificity phosphatase 10 targets miRNA-363-3p, and its inhibition promotes JNK phosphorylation. The miRNA-363-3p/DUSP10/JNK axis was linked to the inhibition of homologous recombination and DNA repair pathways. An innovative therapy is thought to be the miRNA-363-3p/DUSP10/JNK axis (Zhou et al., 2022). miR-517a controls oxidative stress. miR-517a suppression enhances cleaved caspase-3 expression, Bax/Bcl-2 ratio, ROS and MDA levels, and cell apoptosis while decreasing ERK1/2 phosphorylation, T-AOC levels, SOD activity, cell proliferation, and mitochondrial membrane potential. Lower levels of miR-517a result in the inactivation of the JNK signaling pathway. As a result, melanoma cells experience increased oxidative stress (Tofannin et al., 2011). miR-221 demonstrates the impact of cyclin-dependent kinase inhibitor on the occurrence and progression of cell cycle progression (Sun et al., 2011). miRNA-31 identifies the cell division cycle protein 42 and forms a negative feedback chain for JNK inactivation upon the formation of miR-31/Cdc42/phosphorylated MLK3 (p-MLK3). miR-31 and p-JNK were found in high concentrations in the liver tissues of Drug induced lung injury patients with various causes. miR-31 can inhibit the overactivation of the ROS/JNK/mitochondrial diseased death loop in Acetaminophen-induced DILI hepatocytes, suggesting a new therapeutic potential for JNK overactivation-based liver injury. The triggering of apoptosis was mediated. Lower levels of miR-517a result in the inactivation of the JNK signaling pathway. As a result, melanoma cells experience increased oxidative stress (Tofannin et al., 2011). miR-221 demonstrates the impact of cyclin-dependent kinase inhibitor on the occurrence and progression of cell cycle progression (Sun et al., 2011). miRNA-31 identifies the cell division cycle protein 42 and forms a negative feedback chain for JNK inactivation upon the formation of miR-31/Cdc42/p-MLK3. miR-31 and p-JNK were found in high concentrations in the liver tissues of DILI patients with various causes. miR-31 can inhibit the overactivation of the ROS/JNK/mitochondrial diseased death loop in APAP-induced DILI hepatocytes, suggesting a new therapeutic potential for JNK overactivation-based liver injury. The triggering of apoptosis was mediated. Apoptosis was induced by activating the JNK pathway and using a JNK specific inhibitor, which was found to entirely inhibit miR-10b-induced apoptosis.
In this review, we narrated the upregulation and downregulation of miRNAs that target the components of six signaling pathways activated in TB infection. After a careful review of the bibliography, we observed the upregulation and downregulation of miRNAs, playing an important role in the pathways. Most of the signaling pathways remain active during the TB infection. We mentioned the important roles of pathways and their regulating miRNAs. These miRNAs can be considered as therapeutic targets. Therefore, targeting their expression can modulate the activity of signaling pathways. We reviewed the importance of miRNAs, posttranscriptional regulators that control mRNA stability in signaling pathways. miRNAs can multiply and regulate cellular outcomes in response to various extracellular signals by acting as genetic switches or fine-tuners. Signaling networks, on the other hand, control the stability, biogenesis, and abundance of miRNAs over time by regulating layers of the miRNA biogenesis pathway. The detailed study of the miRNAs regulating the immune-associated pathways is useful for the development of miRNA mimetic/inhibitor molecules. Immune effects induced by miRNA drugs are currently the major challenges of miRNA therapeutics.
KD conceptualized the idea, analyzed the data and wrote the manuscript. DC helped in improving the manuscript. All authors contributed to the article and approved the submitted version.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647636 | 36386242 | Gina Córdoba-David,Jorge García-Giménez,Regiane Cardoso Castelo-Branco,Susana Carrasco,Pablo Cannata,Alberto Ortiz,Adrián M. Ramos | Crosstalk between TBK1/IKKε and the type I interferon pathway contributes to tubulointerstitial inflammation and kidney tubular injury 10.3389/fphar.2022.987979 | 23-09-2022 | TBK1/IKKε,type I interferon,kidney injury,TWEAK,LPS,inflammation,cell death | The type I interferon (TI-IFN) pathway regulates innate immunity, inflammation, and apoptosis during infection. However, the contribution of the TI-IFN pathway or upstream signaling pathways to tubular injury in kidney disease is poorly understood. Upon observing evidence of activation of upstream regulators of the TI-IFN pathway in a transcriptomics analysis of murine kidney tubulointerstitial injury, we have now addressed the impact of the TI-IFN and upstream signaling pathways on kidney tubulointerstitial injury. In cultured tubular cells and kidney tissue, IFNα/β binding to IFNAR activated the TI-IFN pathway and recruited antiviral interferon-stimulated genes (ISG) and NF-κB-associated proinflammatory responses. TWEAK and lipopolysaccharide (LPS) signaled through TBK1/IKKε and IRF3 to activate both ISGs and NF-κB. In addition, TWEAK recruited TLR4 to stimulate TBK1/IKKε-dependent ISG and inflammatory responses. Dual pharmacological inhibition of TBK1/IKKε with amlexanox decreased TWEAK- or LPS-induced ISG and cytokine responses, as well as cell death induced by a complex inflammatory milieu that included TWEAK. TBK1 or IRF3 siRNA prevented the TWEAK-induced ISG and inflammatory gene expression while IKKε siRNA did not. In vivo, kidney IFNAR and IFNβ were increased in murine LPS and folic acid nephrotoxicity while IFNAR was increased in human kidney biopsies with tubulointerstitial damage. Inhibition of TBK1/IKKε with amlexanox or IFNAR neutralization decreased TI-IFN pathway activation and protected from kidney injury induced by folic acid or LPS. In conclusion, TI-IFNs, TWEAK, and LPS engage interrelated proinflammatory and antiviral responses in tubular cells. Moreover, inhibition of TBK1/IKKε with amlexanox, and IFNAR targeting, may protect from tubulointerstitial kidney injury. | Crosstalk between TBK1/IKKε and the type I interferon pathway contributes to tubulointerstitial inflammation and kidney tubular injury 10.3389/fphar.2022.987979
The type I interferon (TI-IFN) pathway regulates innate immunity, inflammation, and apoptosis during infection. However, the contribution of the TI-IFN pathway or upstream signaling pathways to tubular injury in kidney disease is poorly understood. Upon observing evidence of activation of upstream regulators of the TI-IFN pathway in a transcriptomics analysis of murine kidney tubulointerstitial injury, we have now addressed the impact of the TI-IFN and upstream signaling pathways on kidney tubulointerstitial injury. In cultured tubular cells and kidney tissue, IFNα/β binding to IFNAR activated the TI-IFN pathway and recruited antiviral interferon-stimulated genes (ISG) and NF-κB-associated proinflammatory responses. TWEAK and lipopolysaccharide (LPS) signaled through TBK1/IKKε and IRF3 to activate both ISGs and NF-κB. In addition, TWEAK recruited TLR4 to stimulate TBK1/IKKε-dependent ISG and inflammatory responses. Dual pharmacological inhibition of TBK1/IKKε with amlexanox decreased TWEAK- or LPS-induced ISG and cytokine responses, as well as cell death induced by a complex inflammatory milieu that included TWEAK. TBK1 or IRF3 siRNA prevented the TWEAK-induced ISG and inflammatory gene expression while IKKε siRNA did not. In vivo, kidney IFNAR and IFNβ were increased in murine LPS and folic acid nephrotoxicity while IFNAR was increased in human kidney biopsies with tubulointerstitial damage. Inhibition of TBK1/IKKε with amlexanox or IFNAR neutralization decreased TI-IFN pathway activation and protected from kidney injury induced by folic acid or LPS. In conclusion, TI-IFNs, TWEAK, and LPS engage interrelated proinflammatory and antiviral responses in tubular cells. Moreover, inhibition of TBK1/IKKε with amlexanox, and IFNAR targeting, may protect from tubulointerstitial kidney injury.
Aberrant activation of innate immunity may lead to maladaptive inflammation and tissue damage during infection and even under sterile pathological conditions. Type I-interferons (TI-IFNs), i.e., interferon-alpha (IFNα) and interferon-beta (IFNβ), are immunomodulatory cytokines typically involved in antiviral responses, which also modulate bacterial or fungal infections (Bogdan et al., 2004; McNab et al., 2015). TI-IFN responses are recruited by structurally conserved pathogen-associated molecular patterns (PAMPs) or host-derived damage-associated molecular patterns (DAMPs). Both PAMPs and DAMPs activate cytoplasmic (RIG, AIM2, cGAS), or membrane (TLR family) receptors which phosphorylate the TBK1/IKKε tandem of noncanonical IκB kinases (IKKs) leading to phosphorylation of interferon regulatory factor (IRFs) transcription factors, such as IRF3 and IRF7, to promote the synthesis of TI-IFNs (Fitzgerald et al., 2003; Hertzog et al., 2003; Honda & Taniguchi, 2006; Crowl et al., 2017). Binding of TI-IFNs to interferon-α/β receptor (IFNAR) induces autophosphorylation of associated JAK proteins (e.g., TYK2, JAK1), leading to STAT1 and STAT2 phosphorylation/activation to induce the transcription of an extensive set of interferon-stimulated genes (ISGs) (Platanias, 2005; Lee & Ashkar, 2018). Overall, the ISG response regulates diverse key cellular processes including apoptosis, autophagy, proliferation, differentiation, and affects the early and late stages of viruses’ life cycle. However, specific roles have only been identified for a few ISGs (Schneider et al., 2014). Crosstalk between the TI-INF pathway and NF-κB (which play a critical role in kidney inflammation) modulates pro-inflammatory transcription and promotes cell survival in human and murine IFNAR-bearing cells, and controls viral infection in vivo (Yang et al., 2000; Sanz et al., 2010b; Rubio et al., 2013; Piaszyk-Borychowska et al., 2019). TBK1 and IKKε also control NF-κB inducers (Shin & Choi, 2019). IRF homodimers or heterodimers cooperate with NF-κB to promote the synthesis of some NF-κB-dependent chemokines, typically Cxcl10, while NF-κB cooperates with IRF3/7 to promote IFNβ synthesis and direct (not dependent on STAT signaling) ISG transcription (Honda & Taniguchi, 2006; Freaney et al., 2013; Schneider et al., 2014; Iwanaszko & Kimmel, 2015). However, whether these interactions contribute to regulating kidney inflammation is unclear, despite the observation that human diseases associated with kidney injury, e.g. viral infections, type I interferonopathies, and autoimmune conditions (e.g., systemic lupus erythematosus) are characterized by enhanced IFN-I signaling (Lodi et al., 2022). Thus, although some reports have highlighted the contribution of T1-IFNs in postischemic kidney injury or lupus nephritis (Freitas et al., 2011; Ding et al., 2021), the regulation and function of molecular routes upstream or downstream of TI-IFN remain mostly unexplored in intrinsic kidney cells and kidney injury, especially in tubular cells and tubulointerstitial disease. TNF-like weak inducer of apoptosis (TWEAK/TNSF12) is a TNF superfamily cytokine that promotes tubular cytokine synthesis, tubular proliferation, and tubulointerstitial inflammation. TWEAK binding to the fibroblast growth factor-inducible 14 (FN14; TNF receptor superfamily member 12a [TNFRSF12a]) receptor activates classical and alternative pathways of NF-κB involving canonical IKKs (IKKα, IKKβ, and IKKγ (NEMO)) [(Poveda et al., 2013)]. Moreover, in the presence of TNFα and IFNγ, TWEAK triggers apoptosis in tubular cells (Justo et al., 2006; Sanz et al., 2011). Therefore, TWEAK/Fn14 axis blockade decreases kidney inflammation and injury in acute kidney injury (AKI) and autosomal dominant polycystic kidney disease (ADPKD) (Sanz et al., 2008; Martin-Sanchez et al., 2018; Cordido et al., 2021). We now describe new molecular pathways activated by TWEAK and LPS involving non-classical IKKs (TBK1 and IKKε) and IRF3, all of which regulate proinflammatory NF-κB activity and trigger the TI-IFN pathway in tubular cells and contribute to kidney inflammation and injury.
Procedures on animals were performed according to the European Community and Animal Research Ethical Committee guidelines and were approved by the IIS-FJD Animal Research Ethical Committee and the Consejería de Medio Ambiente y Ordenación del Territorio, Comunidad de Madrid (PROEX 038/19). The study with human samples complied with ethical precepts formulated in Order SAS 3470/2009 and the Declaration of Helsinki of the World Medical Association on ethical principles for medical research and were approved by the institutional Research Ethical Committee (PIC026-19-FJD). Samples of patients were requested through written informed consent and collected under a biobank regimen.
Murine MCT cells are a well-characterized cell model of kidney tubular epithelium suitable to study molecular mechanisms of kidney injury (Haverty et al., 1988). MCT cells were grown in RPMI 1640 (GIBCO, Grand Island, NY) supplemented with 10% decomplemented fetal bovine serum (DFBS), 2 mM glutamine, 100 U/mL penicillin, and 10 mg/ml streptomycin, in 5% CO2 at 37°C. For experiments, cells were stimulated with 100 ng/ml human TWEAK; 0.01 to 100 mUI/ml IFNα and IFNβ (R&D Systems Inc.); the cytokine mixture made of 100 ng/ml human TWEAK, 30 ng/ml TNFα, and 30 U/ml interferon-γ (IFNγ, PeproTech) or LPS (100 ng/ml, Merck). The following chemical inhibitors were used: 10 μM PF-06700,841 tosylate salt (Sigma-Aldrich, Merck); 50 μM amlexanox; 10 μM Parthenolide, and 2.5 μM IKK16 (MedChemExpress). All the inhibitors were added to cultured cells 1 h before the stimuli. Stock solutions of the stimuli and inhibitors were made according to the specified manufacturers’ instructions. Cells were also treated with the IFNAR neutralizing antibody for 3 h before the stimuli.
Gene transcription was analyzed through quantitative reverse transcription PCR (qRT-qPCR) by using predesigned gene expression assays (TaqMan®, Applied Biosystems-Termofisher Scientific, Waltham, MA, United States). Proteins were assessed by western blot, ELISA, and immunocytochemical-immunofluorescence assays. Standard procedures were applied.
Samples were homogenized in lysis buffer (50 mmol/L Tris, 150 mmol/L NaCl, 2 mmol/L EDTA, 2 mmol/L EGTA, 0.2% Triton X-100, 0.3% NP-40, 0.1 mmol/L PMSF, 25 mmol/L NaF). Proteins were separated by 10% SDS-PAGE under reducing conditions, then blotted onto nitrocellulose membranes. Membrane blockade was accomplished with 5% defatted milk in TBS-T (0.05 mol/L Tris, 0.15 mol/L NaCl, 0.05% Tween 20, pH 7.8). Thereafter, membranes were overnight probed at 4°C with primary antibodies in the same blocking solution or 5% BSA in TBS-T and then incubated with secondary HRP-conjugated antibodies for 1 h at room temperature. The following primary antibodies were used to detect specific proteins of interest: rabbit polyclonal anti-p-STAT1 (Tyr701) (Invitrogen, 44-376G), p-TYK2 (pTyr1054) (Origene, TA333304), and p-IKKε (Ser172) (Sigma Aldrich, 06-1340); rabbit monoclonal anti-p-TBK1/NAK (S172) (D52C2) XP® (Cell Signaling Technology, 1,675,483), TBK1/NAK (E8I3G) (Cell Signaling Technology, 38,066), IKKε (D61F9) XP® (Cell Signalling Technology, 3416), pIRF3 (Ser396) (4D4G) (Cell Signalling Technology, 4,947) and IRF-3 (D83B9) (Cell Signalling Technology, 4,302); monoclonal mouse anti-pIKBα (Santa Cruz, sc-8404). Anti-α-Tubulin (Sigma-Aldrich, MAB374) and anti-GAPDH (Millipore, MAB374) were used to assess protein loading homogeneity.
Cells plated onto glass coverslips were fixed in 4% paraformaldehyde and permeabilized in 0.2% Triton X-100/PBS, washed in PBS, and overnight incubated with polyclonal rabbit anti-p-IRF3 (pSer396) antibody (1:50, Sigma-Aldrich, SAB4504031 or anti-p65 (1:100; Santa Cruz Biotechnology, sc-8008) followed by Alexa 488-conjugated secondary antibody (1:300; Invitrogen) (3). Nuclei were counterstained with DAPI.
Ccl5 expression levels in the supernatants of cultured cells subjected to proinflammatory stimulation were assessed by ELISA (DuoSet ELISA Kit, R&D Systems, Minneapolis, MN) according to the manufacturers’ instructions.
MCT cells were grown in six-well plates and transfected with a mixture of a set of three specific siRNA for TBK1 (75 nM), IKKε (75 nM), or IRF3 (40 nM) (Stealth RNAi™, Invitrogen-TermoFisher Scientific, MA) and Lipofectamine RNAiMAX Transfection Reagent (Invitrogen) made in Opti-MEM I Reduced Serum Medium. After 18 h, cells were washed and cultured for another 48 h in a complete medium containing 10% BSA, and finally serum-deprived for 24 h before stimulation with TWEAK, TTI or z/TTI to evaluate gene mRNA expression and cell death. A siRNA negative control (Stealth RNAi™, Invitrogen) with the same GC content of specific siRNA oligonucleotides was used as a control. After transfection, cells were stimulated at time points when protein expression was reduced by approximately 90%.
For assessment of the overall death rate, cells were washed with PBS following stimulation and then incubated with 0.5 mg/ml MTT (Sigma, Merck) for 1 h at 37°C to detect changes in the metabolic activity. After this step, the MTT solution was withdrawn, and cells were allowed to air dry. Finally, deposits of reduced MTT were dissolved with DMSO, and their absorbance was read at 570 nm. In vivo cell death was assessed by a TUNEL assay performed in 3 µm-thick sections of paraffin-embedded renal tissue (In Situ Cell Death Detection Kit, Fluorescein, Roche Applied Science), according to the manufacturer’s protocol. TUNEL positive cells were counted in 10 randomly chosen fields with a fluorescent microscope.
Models were developed in wild-type0-12-week-old C57BL/6 mice (Charles River Chatillon-Sur-Charlaronne, France) or in TLR4 −/− mice from the same genetic background (Dr S Akira’s laboratory, Osaka University, Japan and generously provided by Dr C. Guerri, CIPF, Valencia, Spain). Four different models of kidney inflammation and injury were assessed. 1) Systemic administration of IFNβ was conducted as previously published (Van-Holten et al., 2004) and dose (0.5 μg/mouse, IP) was established in dose-response test models evaluating renal inflammation. At the times chosen for the experiments (4 and 24 h), recombinant mouse IFNβ (R&D Systems, McKinley Place, MN) caused no tubular cell death or loss of renal function. 2) The TWEAK murine model was previously standardized in our laboratory. It displays renal inflammation originating from classical and alternative activation of NF-κB (Sanz et al., 2010a). Wild-type or TLR4 −/− mice were challenged with recombinant human TWEAK (Merck, Darmstadt, Germany) and renal tissue was analyzed after 24 h. Additionally, some mice also received a second intervention: parthenolide (MedChemExpress, Monmouth Junction, NJ) (70 μg/mouse, IP) or amlexanox (Biorbyt, Cambridge, UK) (50 mg/kg) to inhibit NF-κB or TBK1/IKKε kinases respectively before TWEAK dosage. 3) Endotoxemia-induced kidney inflammation was assessed 24 h after the administration of 5 mg/kg IP bacterial lipopolysaccharide (LPS, Sigma) (LPS nephropathy, LPSN). The dose was chosen on the basis that it may induce a feasible inflammatory reaction and injury in the kidney without hemodynamic compromise and mortality (Panzer et al., 2009; Hato et al., 2019). TBK1/IKKε was inhibited with 50 mg/kg (PO) amlexanox whereas inhibition of the TI-IFN pathway was accomplished with 0.5 mg/kg/day (IP) neutralizing anti-IFNAR antibody (MAR1-5A3, Leinco Technologies). Inhibitors were administered 2 h before LPS. Rat IgG1 (MAB005, R&D Systems) was used as isotype control. 4) Murine folic acid-induced nephrotoxicity (FAN) is characterized by tubular cell death, leukocyte infiltration, and subsequent tubular regeneration that has been reported in humans (Metz-Kurschel et al., 1990; Martin-Sanchez et al., 2018; Yan, 2021). FAN was induced by a single IP injection of 250 mg/kg folic acid and analyzed at 24-96 h (Martin-Sanchez et al., 2018; Yan, 2021). Inhibition with amlexanox or anti-IFNAR antibody was scheduled 2 h before and 24 after FA injection. Mice were euthanized under anaesthesia with 35 mg/kg ketamine (Ketolar/Pfizer) and 5 mg/kg xylazine (Rompun/Bayer). Blood for serum analytical assessment was drawn from the saphena vein and collected on tubes coated with EDTA. Plasma was obtained by centrifugation (1500 rpm, 5 min). Kidneys were perfused in situ with cold saline before removal. One kidney was snap-frozen in liquid nitrogen for RNA and protein studies and the other fixed and paraffin-embedded for immunohistochemistry. Urea plasma levels were assessed by biochemical methods intended for automatic measurements based on the enzymatic decomposition with urease, then followed by colorimetric detection of the reaction product (Roche/Hitachi cobas® c701/702).
Transcriptomics arrays of kidney tissues from mice 24 h after folic acid or vehicle (n = 3/group) were previously published (González-Guerrero et al., 2018; Fontecha-Barriuso et al., 2019). Transcriptomics were performed at Unidad Genómica Moncloa, Fundación Parque Científico de Madrid, Madrid, Spain. Affymetrix microarray analysis followed the manufacturer’s protocol. Image files were initially obtained through Affymetrix GeneChip® Command Console® Software (AGCC) (Affymetrix, Thermo Fisher Scientific, Santa Clara, CA). Subsequently, robust multichip analysis (RMA) was performed using Affymetrix Expression Console® Software Affymetrix, Thermo Fisher Scientific. Starting from the normalized RNA, a significance analysis of microarrays was performed using the limma package (Babelomics, www.babelomics.org), using a false discovery rate (FDR) of 5% to identify genes that were significantly differentially regulated between the analyzed groups. Canonical pathway enrichment analyses were performed using the public database Reactome (www.reactome.org) (supported by United States National Institutes of Health; Toronto University; European Union and the European Molecular Biology Laboratory) 34,788,843 (Gillespie et al., 2022) and Interferome (www.interferome.org) (Monash Institute of Medical Research, University of Cambridge) (Samarajiwa et al., 2009).
Paraffin-embedded sections were stained using standard histology procedures. Immunostaining was performed in 3 μm thick tissue sections that were deparaffinized and antigen retrieved using the PT Link system (Dako Diagnostics, Barcelona, Spain) with Sodium Citrate Buffer (10 mM) adjusted to pH 6-9, depending on the marker. For colorimetric immunohistochemistry, endogenous peroxidase was blocked and then sections were incubated overnight at 4°C with the following primary antibodies: polyclonal rabbit anti-human CD3 (ready to use; DAKO A0452), anti-human MPO (ready to use; DAKO IS511); monoclonal mouse anti-mouse T-bet/Tbx21 [4B10] (1:200; Abcam ab91109); monoclonal rat anti-mouse F4/80 (1:50, MCA497, Bio-Rad); polyclonal rabbit anti-human IFNAR2 (1:100, LS-B13369, LSBio); polyclonal rabbit anti-human IFN beta (1:400, PA5-20390, TermoFisher). Finally, sections were washed, stained with 3,3′-diaminobenzidine (DAB) as chromogen (Dako, Denmark), counterstained with Carazzi`s hematoxylin, dehydrated, and mounted in DPX medium (Merck). Fluorescent immunohistochemistry was developed in tissue sections that were first permeabilized with 0.2% Triton X-100 for 5 min, then blocked with 4% BSA/10% species-specific serum (the same animal species that animal source of primary antibodies), followed by sequential primary and secondary antibodies incubation for 90 m or 60 m, respectively, and finally, the nuclei stained with DAPI for 5 m. Primary antibodies used were: polyclonal rat anti-mouse F4/80 (1:50, MCA497, Bio-Rad); FLEX monoclonal mouse anti-human CD31 (ready to use, GA610, clone JC70A DAKO); polyclonal rabbit anti-human IFN beta (1:100, PA5-20390, TermoFisher). Primary antibody binding to specific antigens was revealed by using the following secondary antibodies (1:200, Invitrogen): Alexa fluor donkey anti-rat 488 (A21208), goat anti-mouse 633 (A21050), goat anti-rabbit 633 (A21070), and donkey anti-rabbit 488 (A21206). Images were obtained by optical (BX53F2 model, Olympus Spain) or confocal (SP5, Leica Microsystems, Spain) microscopy and quantified with ImageProPlus software (Media Cybernetics, Bethesda, MD). Results are shown as the number of positive cells from 10 randomly chosen fields per kidney (20–4×0 objective) or as the percentage of stained area, considering the total area as 100%. Negative controls include non-specific immunoglobulin and no primary antibody.
Biopsies from AKI patients (n = 6) and healthy kidney tissue from nephrectomy specimens (n = 4) were obtained from the IIS-Fundación Jimenez Diaz Biobank (IIS-Fundación Jimenez Diaz, Madrid, Spain). Clinical characteristics of human samples intended for this study are provided in Supplementary Table S1.
Statistical analysis was performed using GraphPad Prism (Dotmatics, San Diego, CA), expressing results as sem ± sd. Significance (p < 0.05) was assessed by a non-parametric Mann-Whitney test for two independent samples.
Recent emphasis on the association of enhanced IFN-I signaling and kidney injury has focused on glomerular injury (Lodi et al., 2022), without considering associated tubulointerstitial injury. Pathway enrichment analysis in a previously reported transcriptomics dataset of murine folic acid nephropathy (FAN) (González-Guerrero et al., 2018; Fontecha-Barriuso et al., 2019) identified multiple activated upstream regulators of the TI-IFN pathway having an absolute z-score > 2.0 and p-value < 0.05 (Supplementary Table S2). Moreover, 168 of the 865 (19.4%) most upregulated genes (fold change ≥1.5 times; p and FDR <0.05) and 37 of the 333 (11.1%) most downregulated genes (fold change ≥0.5 times; p and FDR <0.05) were classified as interferon regulated genes (IRG) in the Interferome database, based on several datasets of mouse cells challenged with TI-IFNs. Subsequent bibliography-based expression profiling analysis also showed overexpression (p and FDR <0.05) of an ISG mRNA signature (Supplementary Table S3). Hence, kidney transcriptomics identified a direct activity of the TI-IFN pathway on ISG transcription or the propagation of an intricate signaling network resulting from the engagement of this pathway in murine tubulointerstitial injury. The increased gene expression of some of these ISGs, IFNβ, and IFNAR1/2, taken as reporters of the TI-IFN pathway activation, were validated by PCR in mice with both FAN and LPS-induced nephropathy (LPSN) (Figures 1A–C). Moreover, in FAN and LPSN mice, increased renal expression of IFNAR2 was detected by immunohistochemistry in kidney tubules (Figure 1D). By contrast, expression of IFNβ, which remained undetectable in control mice, was markedly increased in LPSN and FAN mice in the tubulointerstitial space (Figure 1E) where its fluorescence signal heavily overlaps with the fluorescence signal of endothelial cells (CD31) but hardly with the fluorescence signal of mononuclear phagocytic cells (F4/80) (Figure 1F). In addition, immunohistochemical IFNAR2 staining increased in tubular cells in patients with tubular injury, supporting the clinical relevance of this finding (Figure 1G).
Whether TI-IFNs modulate inflammatory responses in tubular cells has not been firmly established. Thus, we next explored the impact of IFNβ on cultured murine proximal tubular MCT cells. IFNβ dose-dependently increased the mRNA expression of the canonical ISGs Oasl2 and Usp18, peaking at 6 h and decreasing following 24 h, suggesting TI-IFN pathway activation (Figure 2A). Accordingly, IFNβ activated signaling downstream TI-IFN formation through JAK1, TYK2 and STAT1 phosphorylation (Figure 2B) and the transcription of NF-κB target cytokines like Cxcl10 (also considered an ISG), Ccl2, and Il-6, with similar kinetics to canonical TI-IFN-responsive genes (Figure 2C). Supporting crosstalks between the TI-IFN and NF-κB pathways, IFNβ induced IκBα phosphorylation and the nuclear translocation of NF-κB/RELA (p65) (Figure 2D). IFNAR blockade with a specific anti-IFNAR1/2 neutralizing antibody (Figure 2E) or inhibiting TYK2/JAK1 with PF-06700841 (Figure 2F), dampened canonical ISG and cytokine transcription programs. IFNβ also induced phosphorylation of TBK1, IRF3 nuclear location (Figure 1G), and Ifna/Ifnb1 mRNA expression (Supplementary Figure S1A), reflecting a positive activation loop in which IFNβ promotes the expression of TI-IFNs. Inhibition of both TBK1 and IKKε by amlexanox (Reilly et al., 2013) did not modify Oasl2 mRNA levels, which is consistent with IFNAR signaling directly impacting the ISG response. However, amlexanox downregulated Cxcl10 and Ccl2 mRNA expression, without modifying Il6 mRNA levels (Figure 2H), suggesting that IFNAR signaling engages TBK1/IKKε to promote the expression of NF-κB-dependent genes (Figure 2I). Indeed, IFNβ interacts with other key proinflammatory pathways because pre-exposure to IFNβ potentiated TWEAK-induced cytokine transcription to levels above an additive response (Supplementary Figure S1B). IFNα also elicited the ISG and cytokine responses in cultured tubular cells (Supplementary Figure S1C). Consistently, the IFNAR agonist RO8191 (Konishi et al., 2012) mimicked the strong IFNα/β-induced ISG response and the positive activation loop over the NF-κB (Supplementary Figure S1D). Thus, both TI-IFNs share a proinflammatory impact on tubular cells. In vivo, a single dose of IFNβ (Figure 2J) transiently (peak 4 h) increased the kidney expression of ISGs (still mildly increased at 24 h) and proinflammatory cytokines (back to baseline at 24 h) (Figure 2K) resulting in tubulointerstitial inflammation characterized by infiltration by F4/80 + mononuclear phagocytes and CD3+ T lymphocytes that persisted for at least 24 h (Figure 2L). Collectively, these data establish that TI-IFNs promote proinflammatory and antiviral activities in cultured tubular cells and in the kidney in vivo.
We next studied whether proinflammatory stimuli known to cause kidney injury activate the TI-IFN pathway in MCT cells. TWEAK and LPS phosphorylate/activate TBK1, IKKε (Figure 3A), and IRF3 (which translocates into nuclei) (Figure 3B). These results are consistent with previously uncharacterized TWEAK-dependent noncanonical IKKs recruitment, and corroborated TI-IFN production following canonical LPS ligation to TLR4 (Uematsu & Akira, 2007). Furthermore, increases in the IFNβ secretion in TWEAK treatments and in transcription levels of IFNα/β and IFNAR1/IFNAR2 in TWEAK and LPS treatments were found (Figure 3C). TWEAK and LPS also stimulated STAT1 and TYK2 phosphorylation (Figure 3D), i.e. signaling downstream of IFNAR, and correspondingly increased ISG mRNA expression at 6 and 24 h (Figure 3E). The autocrine/paracrine recruitment of the TI-IFN pathway was confirmed by IFNAR blockade (Figure 3F) or TYK2/JAK1 inhibition with PF-06700841 (Supplementary Figure S2A), as both prevented TWEAK- or LPS-induced ISGs transcription (Oasl2) without modifying the cytokine response. Pretreatment with the IKKα/β inhibitor parthenolide limited long-lasting (120 m) TWEAK-induced TBK1 and IKKe phosphorylation (Supplementary Figure S2B) and both parthenolide and a second IKK inhibitor, IKK16, decreased the expression of TBK1/IKKε-dependent ISG genes in cultured tubular cells stimulated with TWEAK and LPS (Figure 3G). These results are consistent with reports of IKKα/β-mediated activation of TBK1/IKKε by NF-κB agonists such as TNFα, IL1β, or LPS, in BMDM and MEFs (Clark et al., 2011). Overall, inflammatory stimuli that cause kidney injury, such as TWEAK and LPS, recruit the canonical NF-κB pathway (IKKα/β) to activate TBK1/IKKε and IRF3-mediated autocrine/paracrine loop of TI-IFN on IFNAR, leading to ISG transcription.
Both TWEAK and LPS activated the TBK1/IKKε-dependent TI-IFN pathway in tubular cells. Therefore, we explored whether TWEAK recruited TLR4, the LPS receptor, to signal through TBK1/IKKε in TWEAK-treated WT and TLR4−/− mice (Figure 4A). In WT mice, TWEAK administration upregulated the kidney mRNA expression of proinflammatory chemokines that are dependent on TBK1/IKKε signaling in cultured tubular cells (i.e., Cxcl10 and Ccl5) (see below, Figure 5B) and this response was milder in TLR4 −/− mice (Figure 4B). In cultured tubular cells, the TLR4 blocker CLI095 prevented the LPS-induced upregulation of cytokine and ISG gene expression, as expected for LPS binding and activation of TLR4 (Figure 4C). To further characterize the in vivo observations in TLR4 −/− mice, we next tested the direct impact of CLI095 on TWEAK-stimulated cultured tubular cells. TLR4 blockade with CLI095 prevented the TWEAK-induced nuclear accumulation of RELA/P65 (Figure 4D) and phosphorylated IRF3 (Figure 4E) and the corresponding increase in Ccl5 and Ifit1 mRNA (6 and 24 h) as well as the persistence of Cxcl10 mRNA upregulation (24 h), without modifying Ccl2 mRNA levels (Figure 4F). Collectively, these data show that TWEAK can transactivate TLR4 to initiate antiviral and NF-κB-mediated transcriptional programs modulated by TBK1/IKKε/IRF3 signaling.
In exploring the participation of TBK1 and IKKε in antiviral and inflammatory responses elicited by proinflammatory stimuli other than IFNβ in tubular cells, amlexanox decreased TWEAK-induced ISG (Figure 5A) and cytokine (Figure 5B) mRNA expression. Consistent with these findings, amlexanox reduced nuclear p-IRF3 levels and NF-κB/p65 location (Figures 5C,D) and decreased p-IκBα levels (Figure 5E). Amlexanox also diminished the LPS-induced gene expression of ISGs and Ccl5 with a time course like TWEAK (Figures 5A,B), whereas Cxcl10 and Ccl2 downregulation was observed by 24 h (Figure 5F). In TWEAK-injected mice (Figure 5G), amlexanox also reduced kidney ISG and cytokine mRNA, consistent with cell culture observations (Figure 5H). In summary, inflammatory mediators such as TWEAK and LPS recruit TBK1/IKKε in tubular cells through pathways that include canonical IKKs and TLR4. Thereby, TBK1/IKKε represents a signaling node located at the crossroad between inflammation and the antiviral response (Figure 5I). Finally, we dissected the functional relevance of each component of the TBK1/IKKε/IRF3 pathway for TWEAK-induced NF-κB and antiviral response activation. Gene silencing reduced TBK1, IKKε, and IRF3 protein levels in tubular cells (≈90% at 48 h) (Supplementary Figure S3A). After further culture for 6 h, TBK1 silencing decreased the basal and TWEAK-induced mRNA expression of ISGs (Figure 6A), and Ccl5 mRNA expression (Figure 6B) and secretion (Supplementary Figure S3B), but it did not modify Cxcl10, Ccl2, and Il-6 mRNA induced responses (Figure 6B). By contrast, together with Ccl5, they were suppressed at 24 h (Figure 6C), suggesting that recruitment of TBK1 promotes the persistence of TWEAK-induced inflammatory responses over time. Combining TBK1 silencing with amlexanox (Figure 6B) or the simultaneous silencing of TBK1 and IKKε thoroughly inhibited the TWEAK-induced mRNA expression of cytokines at 6 h (Figure 6D) suggesting that IKKε is required for early TWEAK-induced NF-κB activation after TBK1 engagement. However, IKKε silencing alone did not modify TWEAK-induced ISG or cytokine responses at 6 h (Figure 6E) or 24 h (Supplementary Figure S3C). Hence, coordinated TBK1 and IKKε activation is required for efficient activation of early TWEAK-induced inflammatory responses whereas only TBK1 is required to activate the ISG program. Finally, like for TBK1 (Figure 6B), IRF3 silencing decreased basal and TWEAK-stimulated ISG mRNA expression (Figure 6F) but only the NF-κB-dependent Ccl5 mRNA and protein upregulation at 6 h (Figure 6G, Supplementary Figure S3D), and the expression of Cxcl10, Ccl2, and Il-6 mRNA at 24 h (Figure 6H), suggesting that the TBK1 axis contributes to the persistence of TWEAK-induced inflammatory responses over time.
In vivo, cells are simultaneously exposed to multiple proinflammatory cytokines that may amplify tubular cell death. A cell microenvironment combining TWEAK, TNFα, and IFNγ (TTI) triggers both an inflammatory response and tubular apoptosis (Justo et al., 2006). Like TWEAK alone, TTI increased cytokine and ISG mRNA levels in cultured tubular cells and this was prevented by inhibition of TBK1/IKKε by amlexanox (Figure 7A). Amlexanox also decreased tubular stress and injury markers Kim-1 and Ngal (Figure 7B) and cell death (Figure 7C) elicited by TTI, suggesting a role of TBK1/IKKε in driving inflammation-induced tubular cell death that is independent of IFNAR, as IFNAR blockade did not prevent TTI-induced cell death (Figure 7D). Altogether, these results support a role for TBK1/IKKε beyond mediating inflammation in also modulating cell death in tubular cells immersed in an inflammatory environment (Figure 7E).
Based on the differential expression of genes related to the TI-IFN pathway in experimental and human kidney injury and the impact of TI-IFNs pathways on tubular cell biology, we addressed the impact of inhibiting TBK1/IKKε signaling or blocking IFNAR in sterile kidney tubulointerstitial inflammation and injury. LPS administration to mice causes endotoxemia and nephrotoxicity. IFNAR blockade by anti-IFNAR antibodies (Figure 8A) decreased the LPS-induced upregulation of mRNA encoding several ISG, including Cxcl10, which targets Th1 lymphocytes and is involved in sepsis-associated kidney injury and other kidney disease conditions (Herzig et al., 2014; Gao et al., 2020) (Figure 8B). IFNAR blockade also decreased the interstitial infiltration by F4/80 + mononuclear phagocytes (Figure 8C), MPO + immune cells and Th1 (T-BET+) lymphocytes (Supplementary Figure S4), and restricted tubular cell death as assessed by TUNEL (Figure 8D). These results suggest that the TI-IFN pathway contributes to LPS-induced kidney inflammation and tubular cell death. We next inhibited TBK1 and IKKε with amlexanox (Figure 8A). Amlexanox protected from kidney dysfunction (Figure 8E), kidney infiltration by F4/80 + cells (Figure 8F), and cell death (Figure 8G). In murine folic acid-induced nephropathy (FAN) (Figure 9A), IFNAR neutralization did not preserve renal function assessed by plasma urea levels (FAN: 245.9 ± 43.3 mg/ml; FAN/IFNAR-Ab: 274.0 ± 25.6 mg/dl; p = ns) but did decrease kidney cytokine and ISG mRNA expression (Figure 9B) and kidney infiltration by F4/80 + phagocytes (Figure 9C), supporting the contribution of the TI-IFN pathway to kidney inflammation. IFNAR blockade also decreased tubular cell death assessed by TUNEL (Figure 9D). Likewise, treatment with amlexanox (Figure 9A) protected from kidney dysfunction (Figure 9E), kidney infiltration by F4/80+ and MPO + cells (Figures 9F,G), and cell death (Figure 9H). Finally, activating the TBK1/IKKε/IRF3 and TI-IFN pathways from the top with DMXAA (Vadimezan), that mimics signaling of TBK1/IKKε initiated by viral dsDNA or bacterial cyclic dinucleotides (Roberts et al., 2007) increased renal cytokine mRNA expression and aggravated the course of FAN following 72-96 h by reducing mice survival. These results suggest deleterious effects of virus or bacteria-induced TBK1/IKKε/IRF3 and TI-IFN pathways over the course of renal injury (Supplementary Figure S4B). Overall, results from nephrotoxic mouse modeling suggest that upstream (TBK1 and IKKε) and downstream (IFNAR) signals within the TI-IFN pathway promote kidney inflammation and injury.
We have now shown that TI-IFNs and proinflammatory factors involved in kidney injury, such as TWEAK and LPS, engage the TI-IFN pathway and associated noncanonical IKKs to activate NF-κB-dependent inflammation and cell death programs, therefore increasing the severity of inflammatory and nephrotoxic kidney injury. Specifically, amlexanox inhibition of TBK1/IKKε was identified as a novel therapeutic intervention for inflammatory and nephrotoxic tubulointerstitial kidney injury. TI-IFNs are thought to contribute to the pathogenesis of viral or autoimmune glomerulonephritis. Glomerular cells express TI-IFN-target genes and cytokines in response to viral-like nucleic acids, and TI-IFNs induce apoptosis and inflammatory responses in these cells (Flür et al., 2009; Migliorini et al., 2013; Lorenz & Anders, 2015). TI-IFN signaling activates NF-κB to confer viral resistance, promote cell survival, and enhance inflammatory gene expression in nonrenal cells (Yang et al., 2000; Rubio et al., 2013; Piaszyk-Borychowska et al., 2019). In tubular cells, we demonstrated that IFNβ activates IFNAR and TYK2/JAK1 to recruit the NF-κB pathway, in a crosstalk that appears to be conserved for different species and cell lineages (Yang et al., 2000). Thus, the TBK1/IKKε signaling node sits at the crossroads of the TI-IFN and NF-κB signaling pathways and arises as a potential regulator of kidney inflammation and injury triggered by TI-IFNs. As maladaptive inflammation plays a key role in amplifying renal injury, NF-κB or related pathways that activate NF-κB, like the TBK1/IKKε/TI-IFN pathway which we now characterized, may contribute to interferon-related nephropathies, and become therapeutic targets (Sanz et al., 2010b; Markó et al., 2016; Gianassi et al., 2019; Song et al., 2019). Tubular cells responded to TWEAK or LPS by activating the TBK1/IKKε/IRF3 pathway together with autocrine/paracrine TI-IFN signaling and ISG transcription. Mechanistically, TBK1 and IRF3 are required for TI-IFN pathway and ISG activation, whereas NF-κB activation depends on IKKε in immune cells (Balka et al., 2020). In contrast, TBK1 and IRF3 mediated NF-κB-dependent proinflammatory responses in tubular cells while IKKε did not contribute by itself to TWEAK-induced NF-κB and TI-IFN/ISG transcription programs. However, IKKε silencing or its pharmacological targeting with AMX in cells with inactive TBK1 allowed the suppression of the early TWEAK-induced NF-κB proinflammatory gene expression. We also report that TI-IFN-related signaling modulates tubular apoptotic cell death. In cultured cell systems, the coordinated TBK1 and IKKε activities, or TBK1 by itself, inhibited RIPK1-dependent TNF-induced apoptosis or necroptosis (Lafont et al., 2018). Indeed, combined TBK1 and IKKε inactivation, or the loss of TBK1, sensitized to TNF lethality in lethal shock in vivo and the human TBK1 deficiency leads to autoinflammation driven by TNF-induced cell death (Lafont et al., 2018; Taft et al., 2021). Likewise, in tubular cells, the simultaneous inhibition of both TBK1 and IKKε by AMX regulated apoptosis, but, on the contrary, it protected tubular cells from TTI-induced apoptosis in vitro and in vivo kidney injury. Therefore, modulation of cell death by TBK1/IKKε inhibition, either by protecting or promoting it, is cell type- or pathology-specific. Classical NF-κB activation by TWEAK depends on the engagement of the IKK complex and subsequent phosphorylation of IκBα (Poveda et al., 2013). Canonical IKKs mediated TNFα-induced TBK1 and IKKε phosphorylation and activation (Clark et al., 2011), in line with the present observation of TWEAK-induced TBK1/IKKε activation in tubular cells. Indeed, the TBK1/IKKε node was engaged by different families of inflammatory mediators (TWEAK, LPS, IFNβ) that share their convergence at NF-κB activation, resulting in reciprocal modulation between canonical and non-canonical IKKs. In addition, TWEAK also recruited the canonical NF-κB pathway through TLR4 transactivation, thus expanding the spectrum of receptors that are transactivated by TWEAK (Rayego-Mateos et al., 2013). TBK1/IKKε-mediated IKKβ phosphorylation had been already described in mouse MEFs, however, in these cells, IKKβ phosphorylation involved residues resulting in IKKβ inactivation rather than in IKKβ activation (Clark et al., 2011). In summary, as in other cell types, interactions between IKK family members were observed in tubular cells. However, the consequences of these interactions vary in a cell type-specific manner. Evidence supporting a contribution of the TI-IFN pathway to kidney disease is incomplete and most related to immune-mediated glomerular injury, in which the contribution of the TI-IFN pathway to tubulointerstitial disease had been overlooked so far. In lupus patients, a pseudoviral immunity state associated with TI-IFN activation and high expression of an ISG signature is associated with more severe disease and nephropathy (Anders, 2009). In lupus-prone mice, interfering with IFNAR signaling improved nephritis. Moreover, clinical trials with the anti-IFNAR antagonistic monoclonal antibody Anifrolumab improved moderate to severe lupus (Felten et al., 2019). Anifrolumab is also undergoing clinical trials for lupus nephritis (Jayne et al., 2022). IFNAR−/− mice are also protected from antibody-mediated glomerulonephritis or post-ischemic kidney injury (Freitas et al., 2011; Deng et al., 2021). We have now shown that in preclinical models characterized mainly by tubulointerstitial injury induced by LPS or folic acid, IFNAR blockade also decreased interstitial inflammation and tubular cell death, suggesting the existence of an intrarenal TI-IFN autocrine/paracrine loop involving renal resident or infiltrating immune cells. Consistent with this notion, in injured kidneys from mice with FAN and LPSN we identified increased IFNAR expression in the tubular epithelium together with a marked signal of IFNβ in CD31+ endothelial cells. IFNβ production by dermal endothelial cells and macrophages is thought to contribute to COVID-19 skin lesions, whereas IFNα from DCs promoted tubular injury in murine kidney ischemia-reperfusion (Deng et al., 2021; Domizio et al., 2022). Therefore, results in LPSN and FAN identify the endothelium as a key source of IFNβ that may bind to tubular IFNAR to amplify inflammation. Additionally, we have now demonstrated the contribution of TBK1/IKKε to both inflammation and cell death in injured kidneys, identifying drugs such as AMX as potentially kidney protective approaches. TI-IFNs induced by viruses with renal tropism have been suggested to contribute to the development or increased severity of glomerulonephritis via direct TI-IFNs stimulation or secondary to release of proinflammatory or tissue-damaging mediators from glomeruli (Lai & Lai, 2006; Flür et al., 2009; Anders et al., 2010; Agrawal et al., 2021). SARS-CoV-2, a cause of acute kidney and glomerular injury, induces TI-IFNs and ISGs responses in tubular cells (Berthelot & Lioté, 2020; Omer et al., 2021). Nephrotoxicity, such as acute tubular necrosis and rejection, has also been observed following antiviral treatment with TI-IFNs in several conditions, as in kidney transplant recipients (Fabrizi et al., 2013). Recently, biopsy-proven thrombotic microangiopathy and focal glomerulosclerosis were reported in patients with IFNβ-associated nephropathy (Dauvergne et al., 2021). Although not reflected in the abstract, acute tubular necrosis was observed in 69% of the cases, being more frequent than glomerulosclerosis (Dauvergne et al., 2021). Data presented in the present manuscript suggest that tubular injury may be a primary event and not necessarily secondary to injury in other kidney structures. Furthermore, hyperactivation of the TI-IFN pathway and collapsing glomerulopathy has been described in STING-associated vasculopathy with onset in infancy (SAVI), an autoinflammatory disease resulting from gain-of-function TMEM173/Sting mutations (Abid et al., 2020). Again, interstitial inflammation and tubular injury were noted in the biopsy. In conclusion, TI-IFNs elicit intracellular signaling events in kidney tubular cells that interact with those elicited by other inflammatory mediators, thus contributing to tubulointerstitial kidney injury under different clinical scenarios, potentially including interferon therapy or the release of endogenous interferons during infection or sterile tissue injury (Figure 10). Specifically, the TBK1/IKKε node was identified as a druggable therapeutic target in kidney disease that acts as a hub linking diverse inflammatory stimuli with tubular cell death and inflammatory responses. |
PMC9647638 | Zhe Liu,Ming-Zhao Yu,Hao Peng,Ruo-Tao Liu,Thou Lim,Chang-Qing Zhang,Zhen-Zhong Zhu,Xiao-Juan Wei | Decellularized tilapia fish skin: A novel candidate for tendon tissue engineering | 06-11-2022 | Decellularized tilapia fish skin,Tendon tissue engineering,Tendon derived stem cells,Tenonic differentiation,Achilles tendon defect | The poor regenerative ability of injured tendon tissues remains a clinical challenge. However, decellularized extracellular matrix (ECM) combined with stem cells shows promise. In contrast to bovine and porcine ECM, marine-derived decellularized ECM has several advantages; it is easily obtained, poses less biological risk, and is not contraindicated on religious grounds. This study successfully fabricated decellularized tilapia fish skin (DTFS) with copious preserved collagen fibers and natural pore structures. The outer layer is smooth and dense, while the inner layer has a soft structure with a rough surface. After crosslinking with 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC) and N-hydroxysuccinimide (NHS), crosslinked DTFS (C-DTFS) showed improved mechanics in dry and wet conditions. In vitro, the leach liquor of crosslinked DTFS showed no cytotoxicity and promoted migration and tenonic differentiation of tendon-derived stem cells (TDSCs). Meanwhile, TDSCs seeded in the inner surface of DTFS maintained viability, differentiated, and exhibited spreading. Furthermore, cell-seeded scaffolds guided the regeneration of tendon tissue in a rat Achilles tendon defect model. Our results suggest that DTFS combined with TDSCs is a novel and promising therapeutic option for tendon tissue engineering. | Decellularized tilapia fish skin: A novel candidate for tendon tissue engineering
The poor regenerative ability of injured tendon tissues remains a clinical challenge. However, decellularized extracellular matrix (ECM) combined with stem cells shows promise. In contrast to bovine and porcine ECM, marine-derived decellularized ECM has several advantages; it is easily obtained, poses less biological risk, and is not contraindicated on religious grounds. This study successfully fabricated decellularized tilapia fish skin (DTFS) with copious preserved collagen fibers and natural pore structures. The outer layer is smooth and dense, while the inner layer has a soft structure with a rough surface. After crosslinking with 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC) and N-hydroxysuccinimide (NHS), crosslinked DTFS (C-DTFS) showed improved mechanics in dry and wet conditions. In vitro, the leach liquor of crosslinked DTFS showed no cytotoxicity and promoted migration and tenonic differentiation of tendon-derived stem cells (TDSCs). Meanwhile, TDSCs seeded in the inner surface of DTFS maintained viability, differentiated, and exhibited spreading. Furthermore, cell-seeded scaffolds guided the regeneration of tendon tissue in a rat Achilles tendon defect model. Our results suggest that DTFS combined with TDSCs is a novel and promising therapeutic option for tendon tissue engineering.
Achilles tendon rupture is a sports injury commonly encountered in clinical practice [1]. Since tendon tissues are characterized by hypocellularity and a lack of nutrient supply, the regeneration of massive tendon defects remains challenging for surgeons [2]. Furthermore, during the healing process, the injured site might be covered by fibrotic scars and exhibit heterotopic ossification, which significantly affect functional recovery [[3], [4], [5]]. Aberrant repair results in poor mechanical properties and a higher risk of re-rupture. Thus, there is an urgent need for an effective therapeutic strategy. Tissue engineering is a promising alternative to autografts, allografts, xenografts, and synthetic prostheses [6,7]. Tissue engineering involves the application of scaffolds, growth factors, and seed cells [8]. For tendon defect repair, appropriate scaffolds must be used. An ideal scaffold for tendon repair should have the biomimetic structure and mechanical properties of the native tendon, and guide tendon regeneration [9,10]. Type I collagen (Col-I) is the main component of the extracellular matrix (ECM) of native tendon, accounting for ∼65–80% of dry tendon mass and 95% of the total tendon collagen [11,12]. Thus, several collagen-based decellularized scaffolds have been widely applied in tendon tissue engineering [13,14]. These scaffolds are mostly mammalian-derived, including decellularized human, porcine, bovine, and equine tendon sheets, dermis, pericardium, small intestinal submucosa, etc. [[15], [16], [17], [18], [19]]. However, these mammal-derived decellularized ECMs have limitations such as limited donor sources, the potential risk of zoonotic diseases, and ethical and religious issues [20]. By contrast, decellularized marine collagen scaffolds could overcome these disadvantages [21]. Decellularized tilapia fish skin (DTFS) has recently received considerable attention because it is easily obtained and processed, and displays excellent biosafety; moreover, its use is not contraindicated on ethical or religious grounds [20,22]. DTFS has been used successfully for burn treatment, bone regeneration, and wound healing [20,22,23]. Besides, 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide and N-hydroxysuccinimide (EDC-NHS) is one of the most common crosslinking agents for collagen materials, offering a non-cytotoxic and zero length crosslinking option, since it is highly water soluble so that it could be easily and thoroughly removed by repeated post-crosslinking rinsing with water [[24], [25], [26]]. Therefore, we investigated whether crosslinked DTFS is suitable for tendon repair or tendon tissue engineering; it merits investigation based on its biological potential and economic value. Seed cells are also important for tissue engineering, and promote tissue regeneration. Several cell types with potential as sources for tendon repair have bene identified, including tenocytes [27], fibroblasts [28], bone marrow-derived mesenchymal stem cells (BMSCs), adipose-derived stem cells (ADSCs) [29], and tendon-derived stem cells (TDSCs) [30]. TDSCs promote more rapid and complete recovery from tendon injuries due to their high proliferative capacity and strong tenonic differentiation ability [31]. Moreover, TDSCs can be easily obtained from injured tendon tissues during surgery. Combined use of TDSCs and ECM scaffolds is an attractive strategy for tendon repair. This study applied DTFS via a mild decellularization method to preserve the native porous structure and abundant collagen fibers, without immunogenic cellular components. To improve mechanical performance, we used a chemical crosslinker and examined the utility of this scaffold for tenonic differentiation of TDSCs in vitro. After that, the TDSCs loaded DTFS was determined to exhibit strong synergetic effect for the regeneration of critical size tendon defect in a rat model (Scheme 1).
The detailed experimental methods can be found in the Supporting Information file.
As shown in Fig. 1A, the macroscopic views demonstrated complete removal of the pigments after decellularization. Histologically, H&E and DAPI staining confirmed the absence of a nucleus in the DTFS groups. Masson's trichrome staining showed a preserved collagen structure (stained blue). Sirius Red staining and immunofluorescence demonstrated that Col-I accounted for a large proportion of the DTFS. On the other hand, the tissues of outer layers had a tight structure, while the inner layers were relatively soft. Electrophoresis showed that both non-crosslinked decellularized tilapia fish skin (NC-DTFS) and crosslinked DTFS (C-DTFS) had a clear background with limited bands, indicating that cell proteins were eliminated (Fig. 1B). However, the collagen components were preserved (α-, β-, and γ-chains of approximately 110–130, 250, and 350 kDa, respectively). Furthermore, DNA quantification (Fig. 1C) revealed that over 95% of the dsDNA was removed following the decellularization process (fresh skin, 215.25 ± 28.51 ng/mg; NC-DTFS, 10.73 ± 3.58 ng/mg; C-DTFS, 4.90 ± 2.10 ng/mg). Additionally, even after decellularizing and crosslinking process, about 8.35 ± 0.21% of TGFβ1, 29.14 ± 1.53% of TGFβ2, and 26.19 ± 1.08% of TGFβ3 were retained in DTFS, and about 7.41 ± 0.11% of TGFβ1, 15.94 ± 0.78% of TGFβ2, and 23.25 ± 0.72% of TGFβ3 retained in C-DTFS; no significant difference between above two groups. These data indicated a successful decellularization process, and that NC-DTFS and C-DTFS had retained an abundant collagen structure that was not significantly affected by crosslinking. The SEM results (Fig. 2A) revealed a relatively smooth outer surface in each group; however, the inner surface exhibit rough appearance. Meanwhile, the inner surface of decellularized skins showed typical criss-cross alignment of collagen fibers, especially in C-DTFS. Besides, the cross-sectional views agreed with the histological findings; the fibers in the inner layers showed soft structure, while the outer layers displayed a dense structure. Moreover, measurement of the dynamic contact angles revealed that the decellularization process increased the wettability of the skin surface, but crosslinking helped to decreased the hydrophilicity in a moderate level. At the same time, the hydrophilicity of outer surfaces of skins is weaker than inner surface in each group (Fig. 2B and C). Since the cells were scheduled to implanted into the inner surface of skins, AFM was used to systemically evaluate its roughness, and the elastic and adhesive stress. There was a trend toward increasing roughness following decellularization, but this was not statistically significant. However, after crosslinking, the roughness improved significantly (Fig. 2D and E). By applying the nanoindentation AFM function, Young's modulus was obtained with the Hertzian model according to the force-distance curves (Fig. 3A). The force is enhanced following an increase in indentation depth, and the slope of the curve reflects the scaffold “modules” (Fig. 3B). The reconstructed map of Young's modulus indicated that the elastic moduli were decreased after decellularization, but crosslinking reversed this and improved the stiffness of the acellular scaffold (Fig. 3C). The C-DTFC sample showed a relatively homogenous distribution of Young's modulus (Fig. 3C). The order of the adhesive force of these skins was as follows: NC-DTFS > C-DTFS > fresh DTFS (Fig. S1). A universal strength testing machine was used to examine the tensile stress of the different skins (Fig. 3D–F). In the dry condition, the free skin had a maximum tensile stress of 7 MPa at 6% strain. After decellularization, the maximum strength decreased to 4.46 MPa at 38% strain. C-DTFS had a maximal tensile stress of 6.11 MPa at 55% strain. C-DTFS had a similar tensile strength but significantly improved extensibility, in comparison with fresh skins. Interestingly, in the wet condition, the strength in all groups decreased and the strain increased; C-DTFS had the highest maximum tensile stress, of 4.12 MPa at 175% strain; that of fresh skin was 1.9 MPa at 53% strain and that of NC-DTFS was 3.54 MPa at 66% strain. Our findings show that C-DTFS has greater tensile strength and extensibility in wet conditions. Considering the mechanical properties of the Achilles tendon (strength and extensibility both value), we used C-DTFS for following biological assessment and performed proteomics analysis to explore the protein composition. We identified 266 peptide groups and 69 proteins with ≥2 unique peptides. Fig. S2 shows the base peak chromatogram. Functional enrichment analysis provided further valuable information (Fig. 4A). Gene Ontology (GO) analysis indicated that C-DTFS contained mostly ECM structural proteins that participated in translation, energy-related activities (like ATP and GTP), and microtubule-based processes. The protein-protein interaction network plot showed that collagen (Col-1, 5, 6, 12, and 14) and fibronectin were the major components, especially the a-chain of Col-I (Fig. 4B). The proteomics analysis illustrated that the collagen-based extracellular structure performed the central functions of the C-DTFS.
We examined the surface markers of isolated TDSCs. After incubation with the corresponding inductive culture medium, the TDSCs successfully exerted osteogenic, adipogenic, and chondrogenic differentiation abilities (Fig. S3A–B). Flow cytometry showed that the stem cell markers CD29, CD44, and CD90 were highly expressed (all >95%), while CD34 and CD45 were almost absent from isolated cells (Fig. S3C). These data strongly indicated that the isolated cells were TDSCs. The effects of the leachates of C-DTFS on TDSCs were examined. A 25–100% concentration of leachates exerted no significant cytotoxic effects on standard MC3T3-E1 cell lines or primary TDSCs (Fig. 5A). Concentration-dependent improvements in cell migration and tenonic differentiation of TDSCs were seen in the leachate treatment groups (Fig. S4 and Fig. 5B). According to western blot results, C-DTFS leachates upregulated Col-I, SCXA, and TNMD protein levels ((Fig. 5C). For potential upstream signals screening, the mRNA levels of collagen and RGD binding related integrin family and TGF receptors were detected by using qRT-PCR. The results indicated that C-DTFS leachate could enhance the expression of integrin α2 (ITGA2), integrin β1 (ITGB1), and TGFBR2 (Fig. S5). The protein level of ITGA2, ITGB1 and TGFBR2 in combined with the phosphorylation level of Smad2 and Smad3 were upregulated by C-DTFS leachate treatment in a dose-dependent manner (Fig. 5D–E). It indicated that the potential cross-talk of integrin α2/β1 and TGFβ/Smad axis contribute to the pro-tenonic effects of C-DTFS. The results of immunofluorescence staining also displayed that the 100% leachate of C-DTFS enhanced the level of Col-I, SCX and TNMD (Fig. 5F).
The day 1 results of CCK-8 showed about 70% implanting rates of both MC3T3-E1 and TDSCs cultured in C-DTFS comparing with commercial tissue culture plates (TCP). Both two type of cells could proliferate when co-cultured with scaffold, and the difference between C-DTFS and TCP was decreased as culture time goes on (Fig. 6A). In line with CCK-8 finding, GFP signals showed the high cell viability of TDSCs seeded in C-DTFS scaffolds (Fig. 6B). The cells successfully implanted in the scaffold and proliferated from day 1–7. F-actin staining revealed a more spreading appearance of TDSCs in day 7 (Fig. 6B). At the same time, SEM revealed the cell numbers and morphology of the TDSCs seeded in the C-DTFS (Fig. 6C); the results were consistent with those of immunofluorescence. In addition, the WB results showed that the level of tenonic markers (Col-I, SCXA, and TNMD) and the upstream signaling (integrin α2/β1 and TGFβ/Smad axis) were all upregulated in C-DTFS group in comparison with standard TCP (Fig. 6C–F). The C-DTFS induced upregulation of Col-I, SCXA, and TNMD were also verified by immunofluorescence staining (Fig. 6G).
Fig. 7A depicts the rolled C-DTFS implanted into the defect site. The pinhole tear strength test indicated that C-DTFS had a strength of approximately 3 MPa, indicating robustness of the sutured tendon (Fig. 7B). The AFI test conducted 2 weeks postoperatively indicated significant functional improvements in the C-DTFS + TDSCs and autograft-sutured groups than the bare C-DTFS and defect groups. C-DTFS-implanted rats also recovered better than the negative controls, with significant differences seen at 8 weeks postoperatively (Fig. 7C and D). Gross observations revealed thickening and heaviness of tendons in all surgical groups at 4 and 8 weeks comparing with normal tendon (Fig. 7E–G); the C-DTFS + TDSCs group showed less thickening than the C-DTFS and defect groups, and was similar to the suture group, indirectly indicating that C-DTFS + TDSCs can decrease fibrous scarring (Fig. 7F). Only the bare C-DTFS group exhibited more weight in the regenerated tendon than the other three surgery groups (Fig. 7G). Histologically, a longitudinal alignment of collagen fibers is seen in the normal Achilles tendon. At 4 weeks postoperatively, aside from minor disorder of the ECM in the suture group, the operated groups exhibited significant losses of fiber alignment, and clear transition of regenerated tissue. Additionally, the defect group displayed obvious atrophy and disordered tendon remnants. After 8 weeks of healing, all implanted groups showed improved collagen fiber arrangements. However, the C-DTFS + TDSCs exhibited a better outcome than the bare DTFS group, with a denser ECM and better-aligned and more mature collagen fibers (Fig. 7, Fig. 8A). At 8 weeks postoperatively, some cartilaginous tissues were present in the C-DTFS and defect groups (Fig. 8A), indicating potential heterotopic ossification. The polarized light on the Sirius Red stained sections denoted regenerated collagen fibers (Fig. 8B). The red and yellow signals indicated Col-I in the normal control and suture groups. Yellow-colored collagen fibers were also present in the C-DTFS + TDSCs group, especially at 8 weeks postoperatively. In the bare C-DTFS groups, the yellow signal was relatively localized and sporadic, and the remaining tissues were characterized by green-colored fibers associated with type III collagen (the main component of scars in tendon tissues). Notably, the defect group showed green-colored fibers, indicating severe scaring. In addition, IHC staining indicated the protein levels of Col-I, SCXA, and TNMD, and showed that C-DTFS + TDSCs treatment had similar effects to autografts, with significantly higher protein levels seen compared with the bare C-DTFS and defect groups (Fig. 8C and S7A-B). Also, the IHC staining reveled higher level of α-SMA in C-DTFS and defect groups, which indirectly indicated the tendon scar formation; while the TDSCs loaded C-DTFS group showed relative lower expression, which is closed to the suture group and normal tendon (Fig. S7C).
Collagen-based scaffolds are essential for tissue engineering in the musculoskeletal system, and especially for skin, bone, cartilage, tendon, and ligament reconstruction [32,33]. Marine-derived acellular collagen scaffolds are emerging as attractive alternatives to traditional mammal-derived materials because they have similar structures and functions to the ECM, lower the risk of disease transmission, and are less likely to be contraindicated on religious grounds [21]. The scale of the fishery industry has rapidly increased, and approximately 70–85% of all products generated are waste or byproducts, including skin, scales, and skeletons [22,34,35]. Converting these unused resources into products that could aid economic or environmental development is of great interest. In clinical practice, MariGen™ (an omega-3 fatty acid product), developed by an Icelandic biomedical company, is the first FDA-approved acellular fish skin-based treatment for burns and diabetic wounds with proven efficacy [36,37]. Besides, MJ. Lacqua in 2021 reported two cases by wrapped this product around repaired tendon to promote its healing and inhibit peritendinous adhesion [38]. Both patients achieved favorable outcomes and excellent functional recovery (100% according to the standard clinical scoring system). Moreover, magnetic resonance imaging revealed complete absorption of the scaffolds with no peritendinous adhesions. Thus, decellularized fish skin shows promise as a tendon rupture treatment due to its multifunctional bioactivities and biodegradability. Inspired by its clinical potential, this study utilized the skin of Nile tilapia, a more popular seafood, to fabricate DTFS and explore its utility for Achilles tendon repair. We compared NC-DTFS and C-DTFS: electrophoresis and DNA content tests demonstrated successful decellularization, similar protein expression patterns and closed TGFβ contents (Fig. 1B–C). Histological findings and SEM observations showed that both DTFSs had relatively dense outer layers with hydrophobic surfaces, and soft inner layers with relatively hydrophilic surfaces. Based on these structural characteristics, the inner surface was used for cell seeding. The relatively dense and hydrophobic outer surface could help prevent fibrotic scarring due to infiltration and inhibit peritendinous adhesion. However, the smooth and soft surface, and suboptimal interface wettability (too high or too low), were not favorable with respect to cell adhesion in most cases [[39], [40], [41], [42]]. Hence, C-DTFS, with its rough and stiff topography and moderate hydrophilicity, was more suitable for seed cell implantation (Fig. 2, Fig. 3, Fig. 4B–C). C-DTFS was also preferred over NC-DTFS because of the demands imposed on the Achilles tendon by intense exercise. C-DTFS had higher tensile strength and better extensibility in both wet and dry conditions, which would significantly reduce the risk of re-rupture. Moreover, when the skin was pulled in the “dorsal-ventral” direction, the collagen fibers were parallel to the tensile stress and could be fully extended, hence behaving similarly to a real tendon. This is especially important for tendon tissue engineering (Scheme 1). Collagens play vital roles in tendon development and healing, and the natural ECM environment enhances the adhesion and tenonic differentiation of stem cells [[42], [43], [44]]. Our proteomic analysis identified abundant tenonic-related collagens (Col-1, 5, 6, 12, 14) in C-DTFS [45], as well as fibronectin (containing the RGD sequence) (Fig. 4). All collagen molecules have a three alpha helix chains, each of which typically contains at least repeating amino acid sequence, such as Gly-X-Y [46]. The GFOGER sequence in Col-I is responsible for integrin receptor recognition, as well as cell adhesion and proliferation, and exhibits desirable biocompatibility for biological applications [47,48]. On the other hand, previous studies demonstrated that integrins can crosstalk with growth factors (GFs) signals to modulate cell differentiation [49,50]. Among these GFs, TGF-β/Smad signaling serves as a paradigm to understand the interaction among integrin, ECM, and GF function in tenonic differentiation [51,52]. Moreover, acellular tendon ECM can enhance integrin α2β1 expression and trigger crosstalk within the TGFβ1R/Smad axis, thereby promoting the tenonic differentiation of ADSCs [29]. Similarly, in our study the leachate of C-DTFS also promoted tenogenesis via the activation of integrin α2β1, TGFβ1R and Smad2/3, suggesting that collagen-enriched C-DTFS has promise for tendon repair. Of note, with EDC-NHS, the standard 100% concentration of crosslink degree is often defined as a precise molar ratio of 5:2:1 (EDC: NHS: COO−), which could be converted into mass ratio with 1.150:0.276:1 (EDC: NHS: Collagen) [24]. Although it demonstrated that EDC mediated crosslink would impair some integrin-specific binding at high concentration; however, using low EDC concentration (∼10% crosslink degree) did not significantly affect cell adhesion, demonstrating that most part of collagen motif could be retained [25,53]. Previous studies indicated that the stability and mechanics of the resultant scaffold would not be affected when the crosslinking degree was reduced to about 10%, and the post-processed scaffolds could be stable for over 28 days in aqueous condition [54]. This kind of collagen scaffolds crosslinked by low degree of EDC, is suitable for long-term applications. In the current study, 50 mg EDC and 20 mg NHS were used for 1 g of DTFS crosslinking. While it was reported that the collagen content in tilapia skin was about 36% so that it could be conducted that the crosslink concentration is lower than 12%, which is safe for cell adhesion and integrin binding [55]. By contrast, glutaraldehyde (GTA) and genipin are other typical crosslinker used in collagen-based products; however, both of them are non-zero length crosslinker so that they could not be removed thoroughly after crosslinking [24]. GTA possesses high cytotoxicity while genipin is biofriendly but would impart a blue color into materials, which is limited in industrial usage [24]. In clinical practice, the massive tendon defect can not be repaired by simply end-to-end suturing or using autografts due to the length or resource restriction, which commonly need the extra allografts or some biomaterials to handle this challenge. Currently, most of the commercially biological products are mammal-derived collagen products, like acellular small intestine submucosa (SIS), dermis, and pericardium [56]. However, just as we mentioned in introduction section, these mammal-derived decellularized ECMs have limitations such as limited donor sources, the potential risk of zoonotic diseases, and ethical and religious issues [20]. Whereas, the C-DTFS in our study, as a decellularized marine collagen scaffolds, could overcome these disadvantages, which owned favorable structural, mechanical, and biological properties, and is suitable for cell seeding. Although its healing effects could not reach the height of autogenous tendon grafts (suture group) according to our results, its broad resource, excellent biosecurity and pro-tenonic effects are of great value for clinical and market requirements. Nevertheless, some limitations still exist and further improvement should be proposed. First of all, the C-DTFS + TDSCs group showed better recovery than the bare C-DTFS group, which indicated that the C-DTFS is more suitable as a cell-loaded scaffold than cell-free option for treatments. While the resource of TDSCs should firstly be taken into consideration in the future, and establishing stem cell bank might solve this problem. Besides cell seeding, the naturally porous structure of C-DTFS could also allow its use as a drug carrier for additional tenonic agents like GFs, platelet-rich plasma, and other pharmacological agents, as an off-the-shelf product. Moreover, according to the in vivo histological findings, the scaffolds were not degraded in rat model until 2 months; therefore, longer observation time and large animal model should be performed in the future. Then, the degradation speed should also be adjusted by modulating the crosslinking degree. What's more, although many fish-derived antimicrobial peptides have been reported, native DTFS had limited antibacterial effects [57,58]. Coating the outer surface of C-DTFS with antibacterial material might be useful for preventing postoperative infection. Additionally, collagen fibers could be extracted to fabricate more homogenous materials via electrospinning or 3D printing. However, mechanical strength might be lost, and further chemical modification or an organic/inorganic hybrid strategy might be necessary.
We created a crosslinked acellular dermal matrix scaffold from Nile tilapia fish skin and described its properties. The results showed that C-DTFS has a natural ECM structure with abundant collagen, and possesses favorable mechanical strength, biocompatibility, and tenonic capacity. C-DTFS is a suitable scaffold for TDSC seeding and aided cell proliferation. More importantly, TDSC-implanted C-DTFS in a rat Achilles defect model promoted the orderly generation of collagen fibers, induced the expression of biological factors important for tendon repair, and decreased the risk of fibrotic scar formation and heterotopic ossification. The process described herein could easily be industrialized and shows great promise for clinical tendon repair applications.
This work was supported by the 10.13039/501100001809National Natural Science Foundation of China (82172400).
Zhe Liu: Conceptualization, Methodology, Investigation, Formal analysis, Writing - original draft, Visualization. Ming-Zhao Yu: Conceptualization, Methodology, Investigation, Formal analysis. Hao Peng: Methodology, Formal analysis, Conceptualization, Software. Ruo-Tao Liu: Formal analysis, Software, Validation. Thou Lim: Conceptualization, Methodology, Investigation, Visualization, Software. Chang-Qing Zhang: Conceptualization, Supervision, Project administration, Resources. Zhen-Zhong Zhu: Methodology, Writing-review & editing, Supervision, Funding acquisition. Xiao-Juan Wei: Conceptualization, Writing-review & editing, Data curation.
The use of animals in these experiments was in accordance with the Interdisciplinary Principles and Guidelines for the Use of Animals in Research, Testing, and Education. The welfare of the experimental animals was prioritized, and all animal experiments were approved by the Animal Care Committee of Shanghai Jiao Tong University Affiliated Sixth People's Hospital at the Shanghai Jiao Tong University School of Medicine and followed the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. |
|
PMC9647639 | 36386685 | Chaosheng Liao,Xiaolong Tang,Maoya Li,Guangrou Lu,Xiaokang Huang,Lin Li,Mingjie Zhang,Yixiao Xie,Chao Chen,Ping Li | Effect of lactic acid bacteria, yeast, and their mixture on the chemical composition, fermentation quality, and bacterial community of cellulase-treated Pennisetum sinese silage | 27-10-2022 | biological pretreatment,lactic acid bacteria,yeast,bacterial community,cellulose,Pennisetum sinese | The present study investigated the effects of Lentilactobacillus buchneri, Saccharomyces cerevisiae, and a mixture of the two on the cellulose degradation and microbial community of cellulase-treated Pennisetum sinese (CTPS) during biological pretreatment. The CTPS was stored without additives (CK) or with L. buchneri (L), yeast (Y, S. cerevisiae), and their mixture (LY) under anaerobic conditions for 60 days. All inoculants enhanced the anaerobic fermentation of CTPS. In relative to L, inoculations with Y and LY decreased the cellulose level of fermented-CTPS by 8.90 ~ 17.13%. Inoculation with L inhibited the growth of Weissella cibaria during anaerobic storage. However, inoculations with LY increased the relative abundance of the homofermentative bacterium Lactiplantibacillus plantarum by 6.04%. Therefore, inoculating S. cerevisiae reduced the adverse effects of L. buchneri-stimulated fermentation on cellulose degradation by altering the bacterial community during anaerobic storage of P. sinese. This work provides a new insight for the subsequent anaerobic digestion of P. sinese. | Effect of lactic acid bacteria, yeast, and their mixture on the chemical composition, fermentation quality, and bacterial community of cellulase-treated Pennisetum sinese silage
The present study investigated the effects of Lentilactobacillus buchneri, Saccharomyces cerevisiae, and a mixture of the two on the cellulose degradation and microbial community of cellulase-treated Pennisetum sinese (CTPS) during biological pretreatment. The CTPS was stored without additives (CK) or with L. buchneri (L), yeast (Y, S. cerevisiae), and their mixture (LY) under anaerobic conditions for 60 days. All inoculants enhanced the anaerobic fermentation of CTPS. In relative to L, inoculations with Y and LY decreased the cellulose level of fermented-CTPS by 8.90 ~ 17.13%. Inoculation with L inhibited the growth of Weissella cibaria during anaerobic storage. However, inoculations with LY increased the relative abundance of the homofermentative bacterium Lactiplantibacillus plantarum by 6.04%. Therefore, inoculating S. cerevisiae reduced the adverse effects of L. buchneri-stimulated fermentation on cellulose degradation by altering the bacterial community during anaerobic storage of P. sinese. This work provides a new insight for the subsequent anaerobic digestion of P. sinese.
Pennisetum sinese, widely known as King grass, a widely used forage crop in the world, is a fast-growing gramineous grass with large biomass and has recently gained increasing attention as an energy crop (Li et al., 2018a). Anaerobic digestion (AD) is known as a widely used and cost-effective way to convert agriculture biomass into biogas (Wu et al., 2019). However, the majority of P. sinese are used for forage utilization and only a few are used for biomass energy production, which is because the lignocellulose in P. sinese is difficult to be decomposed during AD (Patinvoh et al., 2017). Therefore, pretreatment of P. sinese to improve the efficiency of AD is an essential step. Until now, biological pretreatment of lignocellulose has been extensively investigated, such as enzymatic, fungal, composting and ensiling pretreatment (Wang et al., 2020). Among them, the ensiling pretreatment process is again accompanied by the addition of microorganisms, enzymes. Cellulase enzyme is a biological additive often used to pretreat high-fiber plant materials due to their ability to degrade structural carbohydrates to soluble sugars, which act as substrates for subsequent lactic acid bacteria anaerobic fermentation (Colombatto et al., 2004). However, multiple factors, such as enzyme type, concentration, and activity, application method, target substrate, and attached microorganisms of raw materials, govern cellulase activity (Li et al., 2018b). Moreover, the application of cellulase to P. sinese does not always present desirable efficiency due to the enzyme instability during pretreatment (Moharrery et al., 2009). Therefore, the degradation of cellulose by cellulases under microbial conditions is controversial. However, the role of microorganisms in enzyme-driven fiber degradation has rarely been reported. Exploring the effect of cellulase degradation of cellulose under microbial conditions will help to improve the efficiency of subsequent AD of P. sinese. Yeast and lactic acid bacteria (LAB) inoculants have been used as additives during the ensilage of fresh materials, as they preserve the nutritional qualities of the material effectively. Matano et al. (2013) found that yeast mitigated the irreversible adsorption of cellulase onto crystalline cellulose and increased cellulase activity. It means that the presence of yeast may increase the degradation of cellulose during biological pretreatment. However, the role of yeast in promoting cellulose degradation during biological pretreatment has been rarely explored. Meanwhile, Lentilactobacillus buchneri is a heterotrophicus LAB often used as a bacterial inoculant in ensiling pretreatment due to its ability to produce volatile fatty acids, such as acetate and propionate from lactate (Kung et al., 2018). This conversion increases biogas emissions during AD. Stokes (1992) found that the interaction of LAB with cellulase is antagonistic. This implies that inoculation withe LAB maybe inhibits the degradation of cellulose. However, LAB inoculants are almost invariably applied with enzyme additives, making it difficult to differentiate between bacterial and enzyme-mediated ensiling responses (Xu et al., 2017). Few studies have been conducted to investigate the changes in the fiber fraction during biological pretreatment with a combination of cellulose, yeast, and LAB. Moreover, the fiber composition not just has an impact on the efficiency of AD, but also on the subsequent methane production, as cellulose can be degraded and converted to methane (Fujiwara et al., 2022). Therefore, it is essential to investigate the combined effect of microorganisms and cellulases to improve the AD efficiency of P. sinese. The present study investigated the effects of L. buchneri, Saccharomyces cerevisiae and their mixtures as pretreatment inoculations on the fiber fraction and bacterial community of cellulase-treated P. sinese (CTPS). We hypothesized that the inoculation of L. buchneri might affect cellulase activity and inhibit cell wall degradation in P. sinese while adding S. cerevisiae would undo this effect.
For this study, P. sinese was harvested from the yellow cow breeding base in Fenggang County (27°42′ N, 106°55′ E), Zunyi City, Guizhou Province, China. The P. sinese plants were chopped cm to 1–3 cm in length and randomly divided into four blocks with 25 replicates per treatment. The forage from each block was first pretreated with cellulase (F; 2 × 102 μ/g FM, Shanghai Macklin Biochemical Co., Ltd., Shanghai, China, activity, 50 U/mg) and subsequently treated as follows: (i) no additive (CK); (ii) L. buchneri (L; 105 cfu/g FM, Xi’an Jushengyuan Biotechnology Co., Shaanxi, China); (iii) yeast (Y; S. cerevisiae, 2 × 105 cfu/g FM, Xi’an Jushengyuan Biotechnology Co., Shaanxi, China); and their mixture (LY). The inoculations were diluted with sterile water and sprayed evenly on the cellulase-treated P. sinese (CTPS), and CK treated with equal quantities of sterile water. Then, 100 g of CTPS uniformly mixed with each inoculant was manually loaded into a 500 ml Storage Jar (4 treatments ×25 Storage Jar), connected to a collection bag and sealed the jars after tightly vacuumed with an evacuator (SHZ-III type water Circulating Vacuum Pump, Yarong Biochemical Instrument Factory, Shanghai, China). Five Storage Jar per treatment were opened and sampled after 7, 14, 30, 45, and 60 days of storage. A total of 100 samples (4 treatments × 5 storage periods × 5 replicates) were collected and analyzed for fermentation quality, chemical composition and bacterial community composition during fermentation.
Each storage sample of 10 g was uniformly mixed with 90 ml of sterile water in a laboratory juicer for 1 min and filtered through four layers of gauze. The filtrate was then centrifuged at 4,500 × g for 15 min at 4°C. High-performance liquid chromatography (HPLC) was used to evaluate the concentration of butyric, propionic, acetic, and lactic acids (Li et al., 2019). The method of Broderick and Kang (1980) was used to determine the concentration of ammoniacal nitrogen. A pH meter was used to determine the pH of the sample solution. Each storage sample (70 g) was dried at 65°C for a constant weight to determine the dry matter (DM) content, and then ground through 0.20 mm-mesh sieves for analysis of chemical components. Crude protein (CP) content was determined following the AOAC (1990) method. Neutral detergent fiber (NDF) and acid detergent fiber (ADF) content were determined according to the method described by Van Soest et al. (1991). The method of McDonald et al. (1991) was used to ascertain the content of water-soluble carbohydrates (WSC).
To determine the identities of the species present, we used the CTAB method to extract the total genomic DNA from each storage sample. After purification, DNA samples were diluted to 1 ng ml−1 with sterilized water. We amplified the full-length 16S ribosomal RNA (rRNA) gene using specific, barcoded primers (1514R and 27F; Yan et al., 2019). Polymerase Chain Reaction (PCR) amplification was conducted using TransStart®FastPfu DNA Polymerase (TransGen Biotech Co., Ltd., Beijing, China) and the PCR products purified using the QIAquick Gel Extraction Kit (QIAGEN LLC., Germantown, MD, United States). DNA libraries were generated with the SMRTbell Template Prep Kit (PacBio, Menlo Park, CA, United States) and sequenced using the PacBio Sequel system. In order to annotate taxonomic information and assess phylogenetic relationships with the Silva SSUrRNA Database, we used Novogene Bio-Technology Co., Ltd. (Beijing, China) to process raw sequences. Functional prediction, principal coordinates analysis (PCoA), and alpha diversity of the microbial community were assessed using the NovoMagic platform (Novogene Bio-Technology Co., Ltd., Beijing, China).
Data of changes in chemical composition, microbial population and bacterial community indices during storage was repeatedly compared with Duncan’s test, using the SPSS program version 26.0 (IBM Corp., Armonk, NY, United States). Differences were considered statistically significant only when the probability level was lower than 0.05 (p < 0.05). In addition, Spearman correlation was analyzed among bacterial community compositions and anaerobic fermentation parameters.
Table 1 shows the chemical composition of fresh P. sinese before fermentation. The DM of P. sinese was 20.74%, and the NDF and ADF content were 61.34% DM and 35.22% DM, respectively. The WSC content was 5.65% DM, which is close to the WSC content (5.73% DM) reported by Li et al. (2018a). Typically, the WSC of the fresh material acts as an important substrate for silage fermentation, and a material with a WSC content >5% DM contributes to good silage quality (Cai et al., 1998). Thus, the WSC content of fresh P. sinese was sufficient for silage fermentation. Table 2 shows the chemical composition of CTPS after 60 days of fermentation. The DM content did not show differed significantly between the samples. The degradation of CP was mainly initiated by microbial and plant enzymes. In this experiment, the various CTPS samples after fermentation with different inoculants showed no significant difference in the CP content. This is due to the failure of high pH (>4.2) to inhibit microbial degradation of CP. Meanwhile, L and Y inoculated CTPS had higher NDF and ADF contents than CK (CTPS was not inoculated with microbial fermentation) after 60 days of fermentation, this may be attributed to the consumption of available nutrients stimulating an increase in NDF and ADF during anaerobic fermentation (Li et al., 2022). However, the LY-treated CTPS had the lowest NDF (53.49% DM) and ADF (29.70% DM) content, probably due to the LY inoculum promoting cell wall degradation by cellulase. Exogenous cellulolytic enzymes are usually added to the woody fiber material before ensiling. These cellulases degrade cellulose and produce WSC to facilitate anaerobic fermentation. In CTPS, the L treatment (30.77% DM) and Y treatment (28.03% DM) samples had higher cellulose content than CK treated samples (26.33% DM). We attribute this phenomenon to the different effects of L. buchneri and S. cerevisiae on cellulase. Interestingly, the LY-treated (25.50% DM) samples showed a different result from it. Earlier, Chen et al. (2020) reported that a mixture of Lactiplantibacillus plantarum, L. buchneri and cellulase promoted the degradation of fiber fractions. However, Stokes (1992) found that inoculation with a multispecies homofermentative LAB culture (L. plantarum, Levilactobacillus brevis, Pediococcus acidilactici, Streptococcus cremoris, and Streptococcus diacetylactis) was antagonistic to cellulase. This indicated that different microorganisms have different effects on cellulase. Therefore, we believe that the phenomenon observed in this study may be due to the inoculated S. cerevisiae and L. buchneri affecting the cellulase activity, which further affects the degradation of the fiber fraction. The higher ADL content in the samples treated with LY supports this fact.
We further analyzed the fermentation characteristics of CTPS during storage (Table 3). The interactive effect of pH was not significant during the anaerobic fermentation. However, the pH of the L-treated CTPS was the lowest (4.34) after 60 days of fermentation, which was strongly related to the production of lactic acid by the inoculant L. buchneri. Similarly, Zielinska et al. (2015) reported that L. buchneri inoculation decreases the pH of alfalfa silage. The WSC content is a limiting factor for fermentation. Generally, a minimum WSC content of about 3% DM is necessary to successfully preserve material (Zhang et al., 2010). However, WSC, the primary substrate for microbial growth, decreases gradually with fermentation progress. CK treated CTPS had the fastest decrease in WSC content at 7 days of anaerobic storage (5.65% DM to 3.73% DM), and the predominant inoculants inhibited the consumption of WSC by undesirable bacteria during the anaerobic storage. Guo et al. (2014) found more residual WSC in Lactobacillus-treated and fibrolytic enzyme-treated forages, which was Lactobacillus inhibits the consumption of WSC by undesirable bacteria. Besides, the inoculants delayed the decrease in WSC content of anaerobic fermentation compared with CK. Similarly, Li et al. (2022) also reported that the LAB inoculation delayed the decline in WSC. A high level of ammonia-N (>10% of total N) in sample indicates excessive protein breakdown, usually caused by a slight decrease in pH and/or Clostridium fermentation (Kung et al., 2018). In the present study, ammonia-N increased with the fermentation progress. The lowest ammonia-N level was detected in the Y-treated CTPS compared with the control and other additives, indicating good preservation with S. cerevisiae as the inoculant. This may be the result of the inoculated S. cerevisiae inhibiting the growth of NH3-N producing bacteria, thus reducing the NH3-N content in the silage. However, the mechanism underlying the slow degradation of protein to ammonia in the presence of S. cerevisiae is unclear. Lactic acid is considered responsible for the decrease in pH. In this study, lactic acid increased in all CTPS samples, followed by a decrease after 45 days of anaerobic storage, with a maximum at the subsequent time. The changes in lactic acid concentration explain the fluctuation in pH. Notably, at 7 days, the LY-treated CTPS had the highest lactic acid content (1.77% DM). Earlier, Li et al. (2017) also reported that the combination of LAB and cellulase increased the concentration of lactic acid, while Mu et al. (2020) explained that cellulase indirectly provided LAB fermentable sugars by degrading cellulose, subsequently increasing the lactic acid level. Additionally, Carvalho et al. (2021) found that a few yeasts had the potential to produce amylase, cellulase, and protease. Therefore, we attributed the higher lactic acid level in the LY-treated CTPS to the cellulases. We also believe these cellulases led to the lowest cellulose content in the LY-treated CTPS at 60 days. However, the lactic acid level of the LY-treated CTPS lowered than the L-treated CTPS at 60 days, which may be due to the continuous consumption of lactic acid by inoculated yeast (Dolci et al., 2011). Meanwhile, the acetic acid content gradually increased with increasing fermentation time, consistent with Zhou et al. (2016). Studies have shown that the inoculation of heterologous L. buchneri increases fermentation’s acetic acid level (Huang et al., 2021). However, in the L-treated CTPS after 60 days of anaerobic storage, the acetic acid content was not significantly different from the CK-treated CTPS (p > 0.05), probably due to the higher lactic acid that inhibited acetic acid production (Lin et al., 2021). Meanwhile, the low acetic acid levels in the Y-treated CTPS at 60 days could be attributed to the continuous consumption of acetic acid in other bacteria (Ogunade et al., 2017), which will subsequently burden bioenergy production. Studies have proven the presence of propionic acid bacteria that convert glucose and lactic acid to propionic acid and acetic acid (Li et al., 2022). Therefore, the propionic acid produced after 7 days of fermentation in CTPS may be due to the action of these propionic acid bacteria. Under unfavorable conditions, certain undesirable microorganisms, such as Clostridium, convert lactic acid to butyric acid (Steinbrenner et al., 2019). The butyric acid in the LY and CK-treated CTPS may be due to the fermentation of Clostridium. These results suggest that the LY-treated CTPS increased lactic acid content by enhanced cellulose degradation in pre-fermentation, however, the mechanisms behind this need to be explored.
Furthermore, we analyzed the alpha diversity of bacteria found in CTPS after anaerobic fermentation (Table 4). The coverage value of all samples was above 99%, indicating that sequencing adequately captured most of the bacterial communities. The observed species number increased in CTPS during the 60 days anaerobic fermentation. Similarly, Yan et al. (2019) reported an increase in species after anaerobic fermentation of ryegrass. Shannon’s and Simpson’s diversity indices varied similarly among the treatments. The alpha diversity at 60 days was the lowest in the LY-treated CTPS, probably due to the combined effect of S. cerevisiae and L. buchneri. Usually, a low bacterial community diversity is due to the increased abundance of dominant bacteria (Wayne Polley et al., 2007). Meanwhile, ACE and Chao1 are used to evaluate the richness of the microbial community. The richness indices increased in all samples after 60 days of storage, consistent with Ren et al. (2019), who reported an increase in the bacterial richness during anaerobic fermentation of top sugarcane.
Changes in the bacterial community at the genus and species levels during the anaerobic storages are shown in Figure 1 and Table 5. At the genus level, the relative abundance of Lactiplantibacillus and Levilactobacillus increased with increasing storage time, but that of Weissella gradually decreased. Previous studies have shown that Weissella causes fermentation in the early stages, the predominant bacteria gradually shift to Lactobacillus, which is more tolerant to low pH (Cai et al., 1998). In the present study, the LY-treated CTPS had the highest relative abundance of Weissella (14.04%) after 7 days of anaerobic storage, which was also detected after 60 days of storage. The presence of Weissella at the late fermentation stage may be due to the high pH. In the present study, some Bacillus (relative richness >3.0%) was detected in the CK and LY treated CTPS in pre-fermentation. Bacillus accelerates lignocellulose degradation and promotes the stabilization and resource utilization of compost by secreting enzymes (Niu and Li, 2022). Bacillus probably correlated with the high lactic acid concentration of CK-and LY-treated CTPS in the pre-storage period. In addition, Acinetobacter also showed a high abundance in this study. It has been found that Acinetobacter can utilize acetic acid to survive in an anaerobic environment, and its abundance may increase with increasing acetic acid content (Ogunade et al., 2017). Thus, the acetic acid increase may be responsible for the higher abundance of Acinetobacter, the bacterium causes aerobic spoilage via the oxidation of lactic acid and acetic acid (Dolci et al., 2011). The high abundance of Acinetobacter during fermentation may also be responsible for the elevated pH after 30 days of fermentation. The top three abundant bacteria at the species level were L. plantarum, L. brevis, and Weissella cibaria. However, the study did not detect L. buchneri during storage, probably because the L. buchneri used in this study exerted low competition at the beginning of storage, which was replaced by L. plantarum. The high relative abundance of L. plantarum in storage supported this fact. Generally, L. plantarum promotes rapid fermentation that produces lactic acid, preventing further breakdown of the sugars and proteins (Yan et al., 2019). Thus, the lower relative abundance of L. plantarum in the CK-treated CTPS after 7 days of anaerobic storage explains the WSC content. Additionally, we found that L. brevis increased during storage in each treatment group. The Y treatment significantly increased the abundance of L. brevis at 7 days compared to the CK treatment, which may be due to the inoculated S. cerevisiae enhancing cellulose degradation by affecting the cellulase activity, thus providing more substrate for L. brevis growth. L. brevis is a heterologous fermenting bacterium that produces lactic acid and acetic acid as the primary end products through WSC metabolism (Tohno et al., 2012). The abundance of L. brevis also explains the increase in acetic acid content. Meanwhile, the relative abundance of W. cibaria, a heterologous fermenting bacterium, decreased with storage time in overall treated samples. L. plantarum significantly decreased the relative abundance of Weissella, which is usually outcompeted by Lactobacillus spp. as the pH declines during ensiling (Keshri et al., 2019). Interestingly, the LY-treated CTPS had the highest relative abundance of W. cibaria after storage for 60 days. This dominant Weissella resulted in the lowest alpha diversity in the LY-treated CTPS group at 60 days. Teixeira et al. (2021) reported that W. cibaria could ferment prebiotic fibers, which is related to β-glucosidase activity, suggesting that the W. cibaria during anaerobic storage could promote cellulose degradation. Therefore, we attribute the low cellulose content of the LY-treated CTPS in this study to the degradation of cellulose by W. cibaria. Unfortunately, the activity of cellulase was not determined in the current work. In summary, inoculation with L. buchneir and S. cerevisiae increased the abundance of L. brevis and W. cibaria, which enhanced cellulose degradation.
Figure 2A and Table 6 show the predicted functions of the bacterial community during anaerobic fermentation. Chemoheterotrophy and fermentation are ways in which microorganisms utilize organic matter, and their abundance represent the intensity of microbial activity. In the present, the functional group chemoheterotrophy was most abundant in the bacterial community, followed by fermentation and nitrogen-related cycling. The CTPS samples inoculated with LY, L, and Y demonstrated improved fermentation function at 7, 14, and 30 days, respectively. This improvement in function is attributed to the high relative abundance of W. cibaria (Table 5). Studies have proven that the presence of W. cibaria promotes anaerobic fermentation (Cai et al., 1998). In addition, higher nitrite ammonification, nitrogen respiration, and nitrite respiration were detected in the groups with or without the inoculants, which explains the persistent production of ammonia-N. Finally, Spearman correlations between fermentation parameters and kinetics of the top ten genera and species during the fermentation of CTPS were calculated (Figure 2B). The analysis revealed that the pH was positively correlated with Clostridium and L. plantarum, suggesting that the variation in pH was not due to L. plantarum alone. Meanwhile, Weissella correlated negatively with ammonia-N but positively with WSC, implying that the presence of Weissella could preserve the nutritional quality of the anaerobic fermentation. Lactic acid and acetic acid showed positive correlations with L. brevis. Therefore, we inferred that L. brevis was the main component responsible for producing acetic acid during storage. Meanwhile, Bacillus showed a significant negative correlation with NDF, ADF, and cellulose, indicating the role of Bacillus in cellulose degradation, consistent with the reports on the ability of Bacillus to secrete cellulase (Niu and Li, 2022). Furthermore, to explore the effect of inoculants on the bacterial community we performed an analysis of bacterial PCoA. The PCoA plots illustrated that all inoculation promoted the change in the bacterial community during anaerobic fermentation (Figure 3). The LY-treated CTPS remodeled the bacterial community during storage significantly better than the other samples, indicating that S. cerevisiae resulted in a different effect on L. buchneri-mediated anaerobic fermentation.
This study evaluated the effects of L. buchneri, S. cerevisiae, and their mixtures on cellulase treated P. sinese (CTPS) silage quality and PM microbial community. Among the several additives evaluated. S. cerevisiae appeared to be a potential inoculant, and its co-addition with LAB reduced the cellulose content of CTPS during anaerobic storage by increasing W. cibaria and L. brevis. This study showed that the combined addition of S. cerevisiae and L. buchneri reduced the cellulose content during anaerobic storage of CTPS.
The datasets presenting in this article are deposited in NCBI (https://www.ncbi.nlm.nih.gov/) repository, accession number PRJNA887958.
CL: data curation, formal analysis, visualization, writing—original draft, writing—review and editing. XT, ML, GL, XH, LL, MZ: investigation. YX: investigation, resources. PL: conceptualization, methodology, validation, writing—review and editing, supervision, funding acquisition. CC: project administration, funding acquisition. All authors contributed to the article and approved the submitted version.
This project was supported by the National Key Research and Development Program of China (2021YFD1300300) and the basic scientific research funds of Guizhou Province (2022).
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
PMC9647641 | Yingying Ji,Kai Zheng,Shiming Li,Caili Ren,Ying Shen,Lin Tian,Haohao Zhu,Zhenhe Zhou,Ying Jiang | Insight into the potential role of ferroptosis in neurodegenerative diseases | 27-10-2022 | ferroptosis,iron,Alzheimer's disease,neurodegenerative diseases,treatment | Ferroptosis is a newly discovered way of programmed cell death, mainly caused by the accumulation of iron-dependent lipid peroxides in cells, which is morphologically, biochemically and genetically different from the previously reported apoptosis, necrosis and autophagy. Studies have found that ferroptosis plays a key role in the occurrence and development of neurodegenerative diseases, such as Alzheimer's disease, Parkinson's disease and vascular dementia, which suggest that ferroptosis may be involved in regulating the progression of neurodegenerative diseases. At present, on the underlying mechanism of ferroptosis in neurodegenerative diseases is still unclear, and relevant research is urgently needed to clarify the regulatory mechanism and provide the possibility for the development of agents targeting ferroptosis. This review focused on the regulatory mechanism of ferroptosis and its various effects in neurodegenerative diseases, in order to provide reference for the research on ferroptosis in neurodegenerative diseases. | Insight into the potential role of ferroptosis in neurodegenerative diseases
Ferroptosis is a newly discovered way of programmed cell death, mainly caused by the accumulation of iron-dependent lipid peroxides in cells, which is morphologically, biochemically and genetically different from the previously reported apoptosis, necrosis and autophagy. Studies have found that ferroptosis plays a key role in the occurrence and development of neurodegenerative diseases, such as Alzheimer's disease, Parkinson's disease and vascular dementia, which suggest that ferroptosis may be involved in regulating the progression of neurodegenerative diseases. At present, on the underlying mechanism of ferroptosis in neurodegenerative diseases is still unclear, and relevant research is urgently needed to clarify the regulatory mechanism and provide the possibility for the development of agents targeting ferroptosis. This review focused on the regulatory mechanism of ferroptosis and its various effects in neurodegenerative diseases, in order to provide reference for the research on ferroptosis in neurodegenerative diseases.
Iron is involved in oxygen transport and cellular respiration, DNA synthesis and cell division, cellular metabolism and neurotransmission, which are essential for maintaining the body's function and daily metabolism. The ability of iron to circulate in an oxidative state in the body is fundamental to its biological function, and excess iron can lead to oxidative stress damage to biomolecules, as well as cellular dysfunction. However, with the increase of age, the accumulated iron in the brain will increase the risks of neurodegenerative diseases (Belaidi and Bush, 2016; Eid et al., 2017). Ferroptosis is an iron-dependent, novel cell death mode, which is significantly different from apoptosis, cell necrosis and autophagy. The main mechanism is that under the action of ferrous iron or lipoxygenase, iron catalyzes liposomal peroxidation of highly expressed unsaturated fatty acids on cell membranes, thereby inducing cell death (Dixon et al., 2012; Nikseresht et al., 2019). The morphological features of ferroptosis are mitochondrial atrophy, increased bilayer membrane density, and loss of mitochondrial inner membrane cristae, with the intact cell membrane remaining and the normal size of the nucleus, as well as no chromatin condensation (Alborzinia et al., 2018; Ou et al., 2022). Numerous studies have shown that ferroptosis is also related to a reduction in the expression of glutathione and glutathione peroxidase 4 (GPX4) in the antioxidant system of cells (Zhan et al., 2022; Zhao et al., 2022; Zhu et al., 2022). Lipid peroxides cannot be metabolized by the reduction reaction catalyzed by GPX4, and lipids are oxidized by ferrous iron in Fenton reaction to generate a large amount of reactive oxygen species to promote ferroptosis (Alborzinia et al., 2018; Torii, 2018; He et al., 2020). Therefore, the essence of ferroptosis is the metabolic disorder of intracellular lipid oxides, which are then abnormally metabolized under the catalysis of iron ions to produce a large amount of lipids to destroy the intracellular redox balance and attack biological macromolecules, triggering programmed cell death (Figure 1). In recent years, studies have found that ferroptosis plays an extremely important role in neurodegenerative diseases with a common regulatory mechanism. This review will focus on ferroptosis and neurodegenerative diseases, such as Alzheimer's disease (AD), vascular dementia (VD), Parkinson's disease (PD), Huntington's disease (HD), amyotrophic lateral sclerosis (ALS) and traumatic brain injury (TBI), as well as the potential therapeutic effects of targeting ferroptosis (Table 1).
The neuronal iron metabolism related protein, TfR1 (transferrin receptor protein 1), is highly expressed on the surface of neurons (Giometto et al., 1990). Similar to BMECs, iron enters neurons through clathrin-mediated phagocytosis of holo-Tf/TfR1, and exits endosomes in the form of reduced Fe2+ through DMT1 (Burdo et al., 2001). NTBI (Non-transferrin-bound iron) can also enter neurons independently of Tf in a DMT1 (Divalent metal transporter 1)-dependent manner. Prion protein (PrPC), as the ferroreductase partner of DMT1, mediates the uptake of PrPc/DMT1 in the plasma membrane in the form of a complex of iron ions (Shih et al., 2003; Tripathi et al., 2015). Brain iron deficiency and increased holo-Tf uptake can be found in PrPC knockout mice (Singh, 2014). In the brain, divalent iron ions are normally metabolized in the neuronal cytoplasm, and stored in ferritin in the form of trivalent iron ions. When neurons are iron deficient, ferritin can be degraded by lysosomes, releasing stored iron to meet the normal physiological needs of neurons (Connor et al., 1992; Mancias et al., 2014; Raha et al., 2022). Studies have shown that ferritin in the brain increases with age (Belaidi et al., 2018), which is positively correlated with cognitive dysfunction. The etiology of cognitive dysfunction in the elderly is closely related to the iron overload. Iron homeostasis can also be regulated at the translation level. Iron regulatory protein 2 (IRP2), an RNA binding protein, controls the translation of a group of mRNAs involved in iron homeostasis. In the untranslated regions (UTRs) of genes encoding a variety of iron regulatory molecules (including DMT1 and TFR1), the IRP1 and IRP2 bind to the iron response elements (IREs). In an iron-deficient state, the combination of IRP2 and IREs can maximize intracellular iron levels. When the iron content increases, the extracellular iron regulatory pathway (IRE/IRP system) will be activated to weaken the iron overload state (Rouault, 2006; Sanchez et al., 2007; Hentze et al., 2010). In addition, nuclear receptor coactivator 4 (NCOA4) can degrade ferritin to mediate iron autophagy, and release free iron in the process, which can also lead to an increase in intracellular Fe2+ and ferroptosis (Li W. Y. et al., 2021). Iron-responsive element-binding protein 2 (IREB2) is a regulator of iron metabolism, which can up-regulate the expression of ferritin light chain and ferritin heavy chain in the cytoplasm during iron metabolism, and alleviate erastin-induced ferroptosis (Mishima, 2022). Nuclear factor E2-related factor 2 (Nrf2) can reduce the expression of TfR1, regulate iron metabolism, maintain the balance of intracellular iron homeostasis, and limit the production of reactive oxygen species (ROS), thereby reducing ferroptosis (Li S. W. et al., 2021).
Cystine uptake by the glutamate/cysteine antiporter (System Xc-), including a 12-pass transmembrane protein transporter solute carrier family 7 member 11 (SLC7A11) and a single-channel transmembrane regulatory protein solute carrier family 3 member 2 (SLC3A2), is inhibited in ferroptosis (Dixon et al., 2012). Thus, inhibition of the System Xc- results in the depletion of intracellular cysteine (Ma et al., 2021). Cysteine plays an important role in the biosynthesis of glutathione (GSH). GSH, as a substrate of GPX4, is required for its lipid repair function. GSH depletion through cysteine starvation results in the loss of GPX4 activity, as well as the accumulation of unrepaired lipid peroxides and iron toxicity (Angeli et al., 2014). GPX4 converts reduced glutathione to oxidized glutathione (GSSG) to reduce lipid hydrogen peroxide to the corresponding alcohol or free hydrogen peroxide to water (Gaschler et al., 2018). Selenium (Se) is a key regulator of GPX4 activity. Wild-type GPX4 containing Se can effectively reduce peroxides to the corresponding alcohols, thereby preventing ferroptosis (Ingold et al., 2018). GSH is also a natural ligand for Fe2+ in the labile iron pool (LIP), which is an exchange pool for loosely ligated iron in neurons (Hider and Kong, 2011), and glutathione binds Fe2+ in LIP to prevent iron oxidation, which not onlymaintains Fe2+ solubility but also prevents Fe2+ from acting as a catalyst to generate a potent oxidant, hydroxyl radical, from physiologically available hydrogen. Therefore, direct inhibition of GSH synthesis triggers ferroptosis.
Lipid metabolism is also closely related to ferroptosis. Nitrogen oxides (NOXs) provide a source of accumulation of ROS in erastin-induced iron sickness, and it has been reported to modulate the sensitivity of tumor cells to erastin (Dixon et al., 2012). On the other hand, the production of membrane lipid peroxidation is also a source of ROS, which drives iron toxicity. The abundance and location of polyunsaturated fatty acids (PUFAs) determine the degree of lipid peroxidation that occurs in cells, and lead to ferroptosis. The most susceptible lipids are phospholipids containing polyunsaturated fatty acids (PUFA-PLs), which can lead to subsequent cell death (Doll et al., 2017). Free PUFAs need to be esterified to form membrane phospholipids and oxidized to iron ion signals to synthesize lipid signals, especially phospholipids containing phosphatidylethanolamine (PE) and arachidonic acid or epinephrine moieties (Kagan et al., 2017). In the membrane lipid environment, PUFAs are specifically peroxidized in iron toxicity (Doll et al., 2017; Kagan et al., 2017). There are three main classes of lipid oxidases: cyclooxygenases (cox), cytochrome p450s (CYPs), and lipid oxidases (LOXs), of which LOX enzymes have been found to be most important for ferroptosis. LOXs are a class of non-heme iron-containing enzymes that catalyze the deoxygenation of PUFAs (Shintoku et al., 2017).
In addition to β-amylase deposition and accumulation of intracellular neurofibrillary tangles (NFTs) composed of tau proteins, abnormal deposition of iron in the brain is a common feature of AD. The effects of iron on AD have been attributed to its interaction with AD pathological central proteins (amyloid precursor protein and tau protein) and/or through iron-mediated generation of prooxidative molecules such as hydroxyl radicals. However, the source of iron accumulation in brain pathologically relevant regions and its contribution to AD remain unclear. The potential reason for iron accumulation is that senescent cells within tissues increase with age, and these cells trigger inflammation and contribute to various pathologies associated with aging. The accumulation of iron makes aging tissues susceptible to oxidative stress, leading to cellular dysfunction and ferroptosis. In addition, elevated brain iron levels are associated with AD progression and cognitive decline. Elevated brain iron levels, a hallmark of AD, can be pharmacologically modulated to mitigate the effects of age-related dysregulation of iron balance and improve disease outcomes (Masaldan et al., 2019a). A meta-analysis involving 300 AD cases in 19 studies reported that iron levels were significantly elevated in multiple regions of the cerebral cortex, although iron levels varied across regions and studies (Tao et al., 2014). Iron accumulation may be detrimental, as elevation of iron itself may lead to neurodegeneration (Schneider et al., 2012), possibly by inducing oxidative stress and ferroptosis (Stockwell et al., 2017). High brain iron levels, cerebrospinal fluid ferritin (Ayton et al., 2015, 2017a), and quantitative susceptibility maps have been found to have the potential to predict AD clinical severity and cognitive decline (Ayton et al., 2017b). The relationship between postmortem brain iron levels and AD clinical and pathological diagnosis, severity, and rate of cognitive decline in the 12 years preceding death was also investigated in 209 AD patients. It was found that the iron content in the brains of AD patients was significantly increased, and it was significantly related to cognitive function. Therefore, cortical iron may contribute to the deterioration of cognitive function in AD underlying proteinopathies by inducing oxidative stress or ferroptosis, or by being associated with inflammatory responses (Ayton et al., 2020). Another study found that iron deposition in the frontal lobe, parietal lobe, temporal lobe, caudate nucleus, putamen, globus pallidus, cingulate cortex, amygdala, and hippocampus of AD patients was higher than that of healthy controls (Tao et al., 2014), and histological differences in the intensity of iron accumulation in the frontal cortex of AD subtypes can be used not only to distinguish sporadic (late-onset) from familial (early-onset) (Bulk et al., 2018a), but also to correlate with disease severity (van Duijn et al., 2017; Bulk et al., 2018b). The accumulation of iron has been proven to accelerate the deposition of senile plaques and the generation of neurofibrillary tangles (Becerril-Ortega et al., 2014; Kim et al., 2018). Autopsy evidence and MRI analysis provide evidence that there was substantial iron deposition not only in senile plaques (James et al., 2017), but also at sites of cortical tau protein accumulation (Spotorno et al., 2020), suggesting a potential interaction of iron with senile plaques and neurofibrillary tangles. Perturbation of iron homeostasis is one of the key factors in Aβ deposition. High intracellular iron concentration enhances the IRP/IRE interaction and induces upregulation of APP. The enzymes that cleave APP are called α- and β-secretases, which are tightly balanced and regulated by furin (Silvestri and Camaschella, 2008; Guillemot et al., 2013). In the presence of iron excess, more β-secretase is activated when α-secretase is inhibited by furin injury (Silvestri and Camaschella, 2008). Up-regulated APP is cleaved by more β-secretase Aβ40/42, accelerating Aβ deposition. At the same time, the application can no longer assist FPN1, resulting in impaired iron efflux and aggravated iron deposition (Ward et al., 2014). It was suggested that in the absence of redox metals, Aß is nontoxic, and the aggregation of Aß requires the participation of metals (Li et al., 2013; Belaidi and Bush, 2016). Soluble Aß binds to Fe3+ when extracellular iron increases, removing excess iron, but it is difficult to separate after interaction. Aß can promote the reduction of Fe3+ to Fe2+, and ROS released in the process make Aß easily and rapidly to deposit and form more senile plaques (Ha et al., 2007). The interaction of iron with APP and Aß greatly increased the rate and extent of senile plaque formation (Rottkamp et al., 2001). Therefore, iron deposition could be included in the “Aβ cascade hypothesis” of AD (Peters et al., 2018). Iron can also interact with tau protein. Decreased soluble tau protein in the brains of AD patients increases cerebral iron deposition by inhibiting FPN1 activity (Lei et al., 2012). Conversely, high-iron diet led to cognitive decline in mice, abnormally increased neuronal tau phosphorylation, and abnormal expression of insulin pathway-related proteins. Insulin supplementation reduces iron-induced tau phosphorylation (Wan et al., 2019), suggesting that iron deposition may lead to tau hyperphosphorylation by interfering with insulin signaling. In vivo studies have found that iron can participate in tau hyperphosphorylation by activating the cyclin-dependent kinase 5 (CDK5)/P25 complex and glycogen synthase kinase 3β (GSK-3β) (Guo et al., 2013). Excessive intracellular Fe2+-induced generation of oxygen free radicals can also promote tau hyperphosphorylation by activating extracellular signal-regulated kinase 1/2 (Erk1/2) or mitogen-activated protein kinase (MAPK) signaling pathways (Chan and Shea, 2006; Munoz et al., 2006). Glial activation and neuroinflammation have been shown to be prominent features of AD pathology (Newcombe et al., 2018; Leng and Edison, 2021). Microglia are highly responsive cells that respond to increased iron levels in the brain. When iron levels are elevated in the brain, microglia are activated (Meng et al., 2017) with increased volume and decreased length (Rathnasamy et al., 2013; Donley et al., 2021). Iron may activate microglia via nuclear factor-kb (NF-KB)-mediated pro-inflammatory cytokines (Meng et al., 2017). Upon activation, more ferritin will be expressed to remove extracellular iron (Streit et al., 2022), leading to intracellular iron retention (Kenkhuis et al., 2021), increased TNFα expression (Holland et al., 2018), and eventual infiltration as a ß-plaque (Peters et al., 2018; Kenkhuis et al., 2021). It can also interact with APP to promote the formation of Aβ (Tsatsanis et al., 2021). Conversely, in an environment with elevated iron levels, Aß formation leads to increased IL-1ß expression in microglia, exacerbating pro-inflammatory effects (Nnah et al., 2020). Astrocytes are highly resistant to metal-induced toxicity in the brain (Kress et al., 2002), serving as a key cell type for maintaining the homeostasis of the extracellular environment and supporting normal neuronal function (Abbott et al., 2006). Under the high-iron environment, the levels of glutathione, catalase, and manganese superoxide dismutase are significantly elevated in astrocytes to counteract oxidative stress (Iwata-Ichikawa et al., 1999; Shih et al., 2003). However, astrocytes were found to be activated by increased glial fibrillary acidic protein (GFAP) (Kress et al., 2002). Activated astrocytes release inflammatory mediators, induce oxidative stress, and promote the formation of Aß and tau tangles, hindering Aß clearance (Dolotov et al., 2022). Abnormal expression of GPX4 mRNA and its protein levels was found in AD patients and mouse brains (Yoo et al., 2012; da Rocha et al., 2018). In glial cells, mild hypoxia can reduce the level of GSH, which is used for GPX4 biosynthesis (Makarov et al., 2006). In a mouse model of AD, GSH expression was reduced in the cortex and positively correlated with cognitive decline (Karelson et al., 2001). GSH levels in the frontal lobe and hippocampus may serve as biomarkers for predicting AD and mild cognitive impairment (Karelson et al., 2001; Ayton et al., 2020). xCT activity determines GSH availability, and subsequent GPX4 activity in the brain (Ashraf et al., 2020). Furthermore, studies have found that most of the proteins involved in ferroptosis can be regulated by Nrf2 (Habib et al., 2015; Lane et al., 2021). The genes of interest included FPN1, GSH and SLC7A11 encoding xCT. The level of Nrf2 in the brain decreases with age, as well as in AD patients (Osama et al., 2020), so the brain of AD patients is more prone to ferroptosis (Habib et al., 2015). GPX4 expression has been reported to be reduced in both AD mouse models and AD patient brains (Ansari and Scheff, 2010; Yoo et al., 2010). GPX4 knockout mice were shown to have significant hippocampal neuronal loss and cognitive impairment (Yoo et al., 2010; Hambright et al., 2017). A diet deficient in vitamin E, an antioxidant with anti-ferroptosis activity, simultaneously results in hippocampal neurodegeneration and worsens behavioral dysfunction; the ferroptosis inhibitor liproxstatin 1 improves cognitive function and neurodegeneration in mice (Hambright et al., 2017). In addition to animal models, autopsy results of AD patients showed down-regulation of GPX4, up-regulation of arachidonic acid 12/15 lipoxygenase (ALOX15), and enhanced lipid peroxidation, and 4-hydroxynonenal (4-HNE) in AD patient brains was elevated (Yoo et al., 2010). 4-HNE has the potential to modify proteins involved in antioxidant and energy metabolism, promoting Aβ deposition and fibrogenesis (Seibt et al., 2019). These results indicate that ferroptosis plays a key role in AD, which can cause neuronal loss and cognitive decline. Therefore, modulating brain iron metabolism and reducing neuronal ferroptosis may be a promising approach for the treatment of AD. Deferoxamine (DFO) and deferiprone (DFP) are commonly used clinical iron chelators. The clinical efficacy of DFO in the treatment of AD is as high as 50%. However, the side effects, such as weight loss and loss of appetite (Mclachlan et al., 1991), limit its clinical application. Besides, DFO is difficult to pass through the blood-brain barrier (BBB) (Ward et al., 1995; Ben Shachar et al., 2004), which could be solved by intranasal administration of DFO nanoparticles (Rassu et al., 2015). Compared to DFO, DFP could cross the BBB and is safer (Gallie and Olivieri, 2019). In a randomized controlled trial, DFP improved neurological scores and iron-related neurological symptoms (Abbruzzese et al., 2011; Klopstock et al., 2019). Quinoline and its derivatives, which could chelate with iron, zinc, and copper, have the potential to improve cognition, reduce Aβ deposition, and promote Aβ degradation in AD animal models (Grossi et al., 2009; Crouch et al., 2011). Clioquinol could downregulate ß and y secretase and APP expression in the brain (Wang et al., 2012), as well as degrade oligomeric tau protein and reduce tau tangles (Lin et al., 2021). Besides, clioquinol has been proven to slow down cognitive decline in patients with severe AD and reduce Aß42 in cerebrospinal fluid (Ritchie et al., 2003). Studies have also found that vitamin E (an antioxidant) can reduce lipid peroxidation in the brain, reduce iron morphology on neurons, and improve cognitive function in GPX4 knockout mice. However, in a randomized clinical trial, vitamin E showed no benefit in patients with AD or mild cognitive impairment (Farina et al., 2017), while it could accelerate cognitive decline (Lloret et al., 2009). Therefore, the application of vitamin E in AD remains questionable, and more clinical trials are needed to determine its effect. Alpha-lipoic acid (a-LA) was also proven to improve cognitive impairment, slow cognitive decline (Fava et al., 2013), block iron overload, lipid peroxidation, and inflammatory responses in AD patients (Zhang et al., 2018). Se is present in various proteins in the body, such as GPX4, and has antioxidant activity. Se-containing compounds may inhibit iron toxicity by upregulating GPX4 and improve cognitive function in AD patients (Gwon et al., 2010; Cardoso et al., 2019). Ferrostatin-1 (Fer-1) is a common ferritin inhibitor and a free radical scavenger, which is much more effective than phenolic antioxidants (Miotto et al., 2020). Fer-1 could alleviate angiotensin II-induced astroglial inflammation and iron degeneration by inhibiting ROS levels and downregulating Nrf2 and GPX4 (Li S. J. et al., 2021). In the treatment of AD, Fer-1 was shown to ameliorate neuronal death and memory impairment in vitro and in vivo (Bao et al., 2021). Hepcidin can reduce iron transport across the BBB and prevent iron overload in the brain. In cultured microvascular endothelial cells, hepcidin significantly inhibits the expression of FPN1, TfR1 and DMT1, and reduce iron uptake and release by neurons (Du et al., 2015). Hepcidin can reduce the iron level of mouse astrocortex and hippocampal neuron glial cells, reduce the formation of Aß plaques, and improve mouse cognitive function (Xu et al., 2020). The recombinant adenoviruses carrying the hepcidin gene could also reduce iron deposition and oxidative stress levels in the brain (Gong et al., 2016). Insamgobonhwan (GBH) can inhibit the impairment of ß-amyloid on cognitive function in vivo, and also inhibit cell death and lipid peroxidation in vitro cells. In addition, GBH restores ferritin GPX4, HO-1. The expression of COX-2 can improve cognitive dysfunction in mice, and it also has certain potential in AD treatment (Yang et al., 2022).
The main cause of VD is chronic cerebral hypoperfusion (CCH) caused by chronic cerebral blood flow (CBF) and a variety of vascular pathologies. These include atherosclerosis, arteriosclerosis, infarcts, white matter (WM) changes, and microbleeds (Calabrese et al., 2016; Kalaria, 2018). Researchers have demonstrated that amino acid metabolism is related to ferroptosis and that CCH can cause neuronal depolarization to release excess glutamate during the pathogenesis of VD, resulting in excitotoxicity, and high levels of glutamate inhibit the function of System Xc-. Glutamate excitotoxicity is also a pathological mechanism of iron toxicity, and iron chelation prevents excitotoxic cell death (Krzyzanowska et al., 2014; Liu et al., 2016; Frank et al., 2021). Nuclear factor erythroid 2 related factor 2 (Nrf2) is a fundamental regulator of cellular antioxidant defense systems, which regulates the expression of multiple antioxidant response element-dependent genes, including NADPH-quinone oxidoreductase 1 (NQO1), heme oxidoreductase 1 (HMOX1), ferritin heavy chain 1 (FTH1), FPN1, GSH, and GPX4 (Kerins and Ooi, 2018; Milkovic et al., 2019; Sarutipaiboon et al., 2020). Studies have shown that the expression level of Nrf2 is directly related to the susceptibility to iron poisoning. Increased expression of Nrf2 inhibits ferroptosis, and decreased expression promotes ferroptosis (Sun Y. R. et al., 2020; Fan et al., 2021; Nishizawa et al., 2022). Studies have shown that, on the one hand, Nrf2 promotes the expression of glutathione and GPX4 strengthens the function of the antioxidant system, and on the other hand, Nrf2 can also reduce intracellular iron accumulation by promoting the expression of ferritin and the simultaneous release of FPN1 storage and export. iron, thereby preventing iron poisoning (Yang et al., 2017; Kasai et al., 2019). The Nrf2 regulatory network plays a fundamental role in different mouse models of cerebral ischemia. Although the expression of Nrf2 is controversial in different studies, the neuroprotective effect of enhanced Nrf2/ARE activation has been demonstrated in different studies (Park et al., 2018; Liu et al., 2019, 2020). At the same time, NRF2 overexpression can improve cognitive dysfunction (Yang et al., 2014; Qi et al., 2018; Mao et al., 2019), suggesting that it may be related to the inhibition of iron poisoning, and the GSH metabolic network is a bridge connecting iron poisoning and VD. CCH can also lead to massive iron deposition. Bilateral common carotid artery occlusion is the most commonly used experimental model for VD. In the study, it was found that CCH leads to iron deposition in the rat brain, and a large amount of iron deposition leads to neuronal death caused by oxidative stress. Among them, the CA1 area has the most iron deposition and neuronal death (Li et al., 2012; Du et al., 2018). Abnormal brain iron and iron ion deposition are closely related to cognitive dysfunction, and iron ion deposition has been confirmed in AD, PD, HD and other neurodegenerative diseases. It plays an important role in sexually transmitted diseases (Chen L. et al., 2019; Thomas et al., 2020; Xu et al., 2020). It has also been shown that a wide range of abnormal iron deposits in the cortex of patients with subcortical ischemic VD, especially in the lateral caudate nucleus, putamen, globus pallidus, and frontal cortex, correlate closely with the severity of cognitive impairment (Liu et al., 2015; Sun C. Y. et al., 2020). Model of cerebral ischemia-reperfusion injury confirmed that iron accumulation in the ischemic precursor is a novel mechanism of stroke injury, leading to neuronal death. Iron chelators attenuate ischemia-reperfusion injury in animal models (Tuo et al., 2017), indicating that iron-induced ferroptosis may be the underlying mechanism of VD neuron loss. The oxidative stress produced by CCH has been proven to be one of the main pathogenic mechanisms leading to VD (Du et al., 2017; Lee et al., 2021), and studies have shown that blood lipid levels in VD patients are significantly higher than those in AD patients, suggesting lipid peroxidation. Having an important impact on the pathophysiology of VD, MDA levels may be a hallmark of VD (Gustaw-Rothenberg et al., 2010). Lipid peroxidation and ROS accumulation are key processes that induce iron toxicity (Dixon and Stockwell, 2014). LOX causes lipid peroxidation by catalyzing polyunsaturated fatty acids in phospholipid membranes, and inhibition of LOX inhibits ferroptosis (Kagan et al., 2017; Doll et al., 2019). During cerebral ischemia, the extensive increase in 12/15-LOX in brain tissue is an important cause of neuronal cell death and neurological damage. Inhibition of 12/15-LOX can reduce neuronal cell death and brain edema, and improve neurological prognosis (Piao et al., 2008; Pallast et al., 2010; Yigitkanli et al., 2017). In addition, NOX also plays an important role in lipid peroxidation. NOX1 expression was increased in hippocampal neurons during CCH, leading to lipid peroxidation and oxidative stress. It is an important cause of hippocampal neuronal degeneration and cognitive impairment (Choi et al., 2014). Lipid peroxidation caused by NOX is also one of the links of ferroptosis. NOX1 inhibitors have different effects on erastin-induced ferroptosis in Calu-1 cells and HT-1080 cells, and have a partial effect on HT-1080 cells (Dixon et al., 2012; Wang et al., 2021). Thiazolidinediones such as rosiglitazone (ROSI), a drug for the treatment of diabetes, can selectively inhibit ACSL4 activity and thereby inhibit ferroptosis (Angeli et al., 2017; Doll et al., 2017). ACSL4 is widely expressed in brain tissue, especially in the hippocampal CA1 region, and the expression of ACSL4 is gradually increased during cerebral ischemia (Kassan et al., 2013; Peng et al., 2021). Rosiglitazone (ROSI) has been shown to reduce lipid peroxidation and oxidative stress injury in hippocampal neurons during CCH, protecting brain function (Sayan-Ozacmak et al., 2012). Multiple studies have shown that long-term use of pioglitazone in patients with insulin-dependent diabetes reduces the risk of dementia in nonpsychiatric patients (Heneka et al., 2015; Lu et al., 2018).
PD is the second most common neurodegenerative disease. It is common in middle-aged and elderly people with neurodegenerative diseases. Some PD patients also have cognitive dysfunction. In the late stage of PD, patients often have severe cognitive dysfunction such as dementia. Abnormal intracranial iron deposition is thought to be one of the pathogenic mechanisms of PD (Langkammer et al., 2016; An et al., 2018; Chen Q. et al., 2019), and iron induces the formation of Lewy bodies through oxidative stress pathway, which aggregates α-synuclein (Takahashi et al., 2007). Li et al. (2018) found that the iron content of the substantia nigra pars compacta in PD patients may gradually increase during the progression of PD and manifest as more significant iron deposition in the middle and late stages of the disease. Relatedly, observation of changes in its iron deposition may serve as a potential marker for monitoring disease progression. Substantia nigra iron deposition is also related to the cognitive function of patients (Liu et al., 2017). The QSM CP value of the left substantia nigra on MRI was negatively correlated with the ADL score. The CP value of the left frontal white matter was negatively correlated with the HAMD score, CP value of left substantia nigra and CP value of left frontal lobe were positively correlated with MOCA score, and CP value of left frontal lobe white matter was positively correlated with MMSE score. SWI reflects abnormal iron deposition in the brain through CP value, which can be used for the diagnosis of PD, but has little significance for disease staging, and can be used to study the mechanism of PD cognitive dysfunction and depression (Xiong et al., 2020). Quantitative analysis of susceptibility-weighted imaging (SWI) found that PD patients with mild cognitive impairment had increased iron concentrations in the globus pallidus and head of the caudate nucleus (Kim et al., 2021). Iron is also widely deposited in the premotor cortex, prefrontal lobe, insula, cerebellum, pons and other parts of PD patients (Acosta-Cabronero et al., 2017; Chen L. et al., 2019), which are all related to apathy and rapid eye movement sleep behavior disorder (RBD) and other non-motor symptoms. Tere was a metabolic disorder of iron in the cerebrospinal fluid of patients with PD combined with apathy and PD combined with RBD (Hu et al., 2015; Wang et al., 2016). Serum iron has also been found to be associated with anxiety in PD patients (Xu et al., 2018). In addition, Masaldan et al. (2019b) found that the abnormal accumulation of iron in PD may be closely related to the changes of iron regulatory proteins. Alpha-synuclein (a-syn), as a key player in tyrosine hydroxylase-dependent dopamine synthesis and other dopamine metabolic processes (Do Van et al., 2016; Belarbi et al., 2017), may play a role in iron regulation (Duce et al., 2010; Zhou and Tan, 2017). In the presence of a copper catalyst, a-syn has ferroreductase potential, which, when combined with Fe3+ and converted to Fe2+, binds to the C-terminus of a-syn (Davies et al., 2011). Iron also increases the formation of non-normal fibers, a major event in PD (Abeyawardhane et al., 2018). Furthermore, GSH was found to be decreased in a mouse model of MPTP (Feng et al., 2014), while GSH depletion enhanced MPP+ toxicity in substantia nigra dopaminergic neurons (Wullner et al., 1996). DFP has also been found to be neuroprotective in patients with early PD (Do Van et al., 2016). The MPP+-induced SH-SY5Y (a commonly used PD model) cell line is not programmed cell death and shares some similarities with iron toxicity: both involve lipid peroxidation and can be inhibited by DIM and Fer−1 suppressed. Results showed that iron chelators not only inhibited iron toxicity, but also protected dopamine neurons from cell death (Abeyawardhane et al., 2018). In a transgenic mouse model of PD, clioquinol was able to prevent the loss of substantia nigra cells due to the ability of clioquinol to chelate iron (Billings et al., 2016). Zeng et al. (2021) found that GPX4 in PD cells significantly decreased and ROS increased significantly after administration of ferric ammonium citrate, which in turn induced ferroptosis and led to neuronal death, while administration of iron chelators could inhibit ferroptosis and protect neurons.
HD is a progressive neurodegenerative disease characterized by rapid involuntary movements and cognitive impairment, ultimately leading to death, due to expansion of CAG repeats in the Huntingtin (HTT). The pathological hallmark of HD is iron accumulation and abnormal levels of glutamate and glutathione (Skouta et al., 2014; Agrawal et al., 2018). It has also been reported that plasma samples from HD patients have lower levels of GSH (Klepac et al., 2007) and lower GPX activity in erythrocytes, so the pathogenesis of HD may be related to ferroptosis. In a mouse model of HD, nitropropionic acid-treated mice also exhibited reduced global (cytoplasmic and mitochondrial) GSH reduction, suppressed hippocampal and cortical glutathione s-transferase (GST) function, and exhibited an HD phenotype (Klivenyi et al., 2000). Although the underlying mechanism by which mutant huntingtin causes neurodegeneration is unclear, the ability of huntingtin to induce oxidative damage has been demonstrated. A study found that Fer-1 treatment at 10 nM, 100 nM, and 1 μM protected neurons labeled with yellow fluorescent protein (YFP) and expressed by biotransfection with a pathogenic repeat (73Q) The huntingtin exon 1 fragment (mN90Q73) induces cell death. The number of medium spiny neurons (msnn) was significantly increased compared to controls (Skouta et al., 2014). The iron chelator DFO was protective and improved cognition in R6/2 mice, a mouse model of HD (Yang et al., 2016). Fer-1 can inhibit oxidative lipid damage and cell death in a cellular model of HD (Skouta et al., 2014).
ALS is a neurodegenerative disease affecting motor neurons in the cortex, spinal cord, and brainstem, and clinically manifests as progressive muscle atrophy and weakness in the extremities and trunk. The pathogenesis of ALS is unknown, and neither is the treatment of the disease. The most common type of dementia associated with ALS is frontotemporal dementia (FTD), a progressive non-AD dementia syndrome characterized by localized frontal and temporal lobe degeneration (Ringbolz and Greene, 2006; Heidler-Gary and Hillis, 2007; Bede et al., 2018; Iridoy et al., 2019). Iron deposition was observed in the spinal cord of a mouse model of ALS (Golko-Perez et al., 2017). Abnormal iron deposits are found in ALS patients, and iron deposits were found in the motor cortex of the patient's brain at autopsy (Kwan et al., 2012). The lipid peroxidation of erythrocytes in ALS patients is significantly increased, while the content of GSH is decreased (Babu et al., 2008). The concentration of glutamate is higher, and the accumulation of glutamate can cause neuronal cell toxicity, which indicates that ferroptosis is directly involved in the pathogenesis of ALS mechanism. At the same time, motor neurons in ALS mice are very sensitive to GPX4 knockout-induced cell death (Conrad et al., 2018). At present, the only approved treatment for ALS in the United States and the European Union is the anti-excitatory amino acid toxicity drug Riluzole, which is a glutamate antagonist, which can inhibit the release of presynaptic glutamate and inhibit nerve endings. It can inhibit the neurotoxicity of excitatory amino acids by inhibiting the neurotoxicity of excitatory amino acids. It has a certain effect on ALS patients with bulbar palsy or limb paralysis as the first symptom, can delay the progression of ALS, and can clearly prolong the survival time and postponement of tracheostomy (Gurney et al., 1998).
TBI is recognized as a disease with high mortality and complex survival, which is the main environmental risk factors for the development of neurodegenerative diseases. Complications of TBI are mainly motor function, cognitive function, and social dysfunction, which cause a serious burden on patients, family members and society, and age is an important factor affecting the prognosis of TBI (Griesbach et al., 2018; Fraser et al., 2019; Ritzel et al., 2019). Iron is considered to be an important agent of secondary injury after traumatic brain injury and can induce peroxidation and inflammation (Ayton and Lei, 2014). Studies have shown that after experimental brain trauma in rats, the production of lipid peroxidation products is significantly enhanced, the consumption of GSH and ascorbic acid is significantly increased (Bayir et al., 2002), and the activity of GPX is decreased (Xu et al., 2014). Elevated levels of 15-HpETE-PE after traumatic brain injury led to ferroptosis in the cerebral cortex and hippocampus, accompanied by increased expression of 15LO2 (a catalyst for the formation of protoferroporphyrin 15-oh-eicosapentaenoic acid) and depletion of GPX4, leading to cognitive impairment, effectively suggesting the possibility of ferroptosis (Wenzel et al., 2017). Iron chelators such as DFO may improve cognitive function after traumatic brain injury (Khalaf et al., 2019). N,N'-bis(2-hydroxybenzyl)ethylenediamine-N,N'-diacetic acid hydrochloride (HBED) is a unique iron chelator that not only crosses the BBB, but also reduces the improvement and recovery of motor impairment and cognitive function after TBI in rats (Khalaf et al., 2019). The ferroptosis inhibitor Liproxstatin 1 can reduce brain edema and BBB permeability caused by TBI, improve motor and learning and memory impairment caused by TBI in rats, and significantly improve anxiety and cognitive function caused by TBI (Xie et al., 2019).
Ferroptosis is a newly discovered form of cell death manifested by iron overload, accumulation of lipid peroxidation, and ROS. The current research preliminarily showed that ferroptosis plays an important role in neurodegenerative diseases. Clinically, ferroptosis can be induced by the following methods and exert a neuroprotective effect: exogenous lipid supplementation promotes lipid peroxidation in cells; inhibition of GPX4 and expression of GSH; construction of nano-drug delivery system to supplement hydrogen peroxide and iron ions to promote Fenton reaction of tumor cells, etc. Therefore, ferroptosis is a potential target for the treatment of neurodegenerative diseases. However, the exploration of ferroptosis still faces many problems to be solved. First, the study of ferroptosis in cognitive dysfunction-related diseases is still in its infancy, and its underlying molecular mechanisms remain unclear. Although iron overload and lipid peroxidation can cause ferroptosis, their involvement in ferroptosis-related regulatory targets such as DMT1, FPN1, or iron uptake proteins needs to be further investigated. Second, there is no specific ferroptosis marker to comprehensively and extensively study its process, and which are the key executive molecules of ferroptosis remain unclear. Finally, it is known that abnormal iron metabolism can cause ferroptosis, and whether other metal elements induce ferroptosis remains to be explored. In addition, ferroptosis is different from other forms of cell death, but these different forms of cell death are not independent of each other. The various forms of cell death are likely to be interconnected and form a network to participate in the regulation of cell death. Studies have found (Sun et al., 2018; Chen et al., 2021) that ferroptosis is closely related to apoptosis, and ferroptosis, autophagy and apoptosis can synergistically promote cancer cell death. Therefore, further studies on the relationship and mechanism between ferroptosis and other known cell death pathways are still needed in the future, which may be helpful for the treatment of cognitive dysfunction-related diseases. In addition, ferroptosis inhibitors contain some traditional ROS scavengers, but compared with traditional ROS scavengers, ferroptosis inhibitors can block iron-catalyzed ROS generation, activate oxidative stress, and induce cell death. It can only clear the accumulated ROS in cells and has a weak inhibitory effect on ferroptosis, and cannot completely block the occurrence of ferroptosis in cells. Various types of ferroptosis inducers and inhibitors have been found, but ferroptosis regulators generally suffer from low bioavailability and adverse reactions. Therefore, screening of ferroptosis-related drugs and traditional Chinese medicines with few adverse effects on normal tissues and high target specificity is crucial for the development of ferroptosis. In conclusion, with the gradual deepening of ferroptosis research, the research on targeted drugs and new drug targets for ferroptosis is of great significance for the prevention and treatment of diseases in the future. At the same time, more and more experimental studies have confirmed the role of ferroptosis in neurodegenerative diseases, which provides more possibilities for the discovery of potential therapeutic drugs and therapeutic targets for neurodegenerative diseases, and also provides help to further explain the pathogenesis of neurodegenerative diseases. However, the research on ferroptosis in neurodegenerative diseases is still in its initial stage, and there are many unexplained problems, and more experiments are needed to deepen its understanding.
YJi and ZZ conceived the study. KZ, SL, CR, YS, LT, and YJia performed literature searching and summary. YJi, HZ, and YJia wrote the manuscript. YJia and ZZ edited the manuscript. All authors contributed to the article and approved the submitted version.
The work was supported by the National Natural Science Foundation of China (No. 82104244), Wuxi Municipal Health Commission (Nos. Q202050, Q202101, Q202167, M202167, and ZH202110), Wuxi Taihu Talent Project (Nos. WXTTP2020008 and WXTTP2021), Wuxi Medical Development Discipline Project (No. FZXK2021012), and Jiangsu Research Hospital Association for Precision Medication (JY202105).
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. |
|
PMC9647642 | Sue Ann Costa Clemens,Ana Keiko Sekine,Fernanda Tovar-Moll,Ralf Clemens | COVID-19 site readiness initiative: Building clinical trial capacity for vaccine efficacy trials in Latin America in response to the pandemic | 08-11-2022 | Capacity building,Clinical trial,Clinical trial site,COVID-19,Latin America,Vaccine trial | Highlights • BMGF launched a trial site readiness grant for COVID-19 clinical development. • Latin America sites recently successfully executed various vaccine efficacy trials. • Within 4 months 21 sites were ready to participate in COVID-19 vaccine studies. • Infrastructure and equipment improvements consumed most of the sites’ budget (81%) • This site readiness initiative may be a blueprint for public health emergencies. | COVID-19 site readiness initiative: Building clinical trial capacity for vaccine efficacy trials in Latin America in response to the pandemic
• BMGF launched a trial site readiness grant for COVID-19 clinical development. • Latin America sites recently successfully executed various vaccine efficacy trials. • Within 4 months 21 sites were ready to participate in COVID-19 vaccine studies. • Infrastructure and equipment improvements consumed most of the sites’ budget (81%) • This site readiness initiative may be a blueprint for public health emergencies.
Randomized clinical trials are considered the gold standard for assessing the safety and efficacy of new drugs and vaccines. Historically, high- and to a lesser extent high-middle income countries (HICs and HMICs) have been the biggest contributors of clinical trial data, but over the past decade many lower middle-income countries (LMICs) have received significant investments in a bid to develop capacity and infrastructure to conduct vaccine trials. Even so, some HMICs and LMICs still struggle to build sustainable capabilities or provide qualified regulatory oversight for conducting vaccine trials and maintain adequate infrastructure, thereby often needing the support of product development partners (PDPs) [1], [2]. In December 2019, a large number of people in Wuhan were reported to have pneumonia of an unknown origin which was subsequently identified as an epidemic due to a new strain of virus called severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2). On 11th March 2020, the World Health Organization (WHO) announced this outbreak as a pandemic (coronavirus disease-2019 [COVID-19]), having spread across 6 continents [3], [4], [5]. The situation demanded immediate evidence-based research, calling for global collaborative efforts in the preparation and conduct of randomized controlled vaccine trials to provide evidence for controlling the pandemic [6]. As of 2nd December 2021, the current number of COVID-19 trials registered at ClinicalTrials.gov stands at 3,939. According to the WHO [7], [8], there were 140 candidate vaccines for COVID-19 under clinical development. The global disease burden COVID-19 stands at 628 million cases and over 18 million deaths [9]. Despite the recent declining trend, the American region has the largest disease burden (37 % of total cases and 44 % of total deaths) with the highest number of new cases and deaths reported from the United States of America (USA), Brazil and Mexico [10]. Latin America has a tradition of vaccine development, having participated in recent years in large efficacy trials including the Rotavirus vaccines, the human papillomavirus vaccines (HPV), the Pneumococcal Conjugate vaccine as well as the two recent Dengue vaccine trials [11], [12], [13], [14], [15]. The region offers attractive development conditions for vaccine clinical trials owing to well-trained physicians and staff especially in infectious diseases, the high acceptance rate of vaccines, the distinct environment and ethnic diversity [16], [17], population density greater than the USA or Europe [18], dense urban areas [19], high-quality standards for conducting clinical trials, strict regulatory guidelines based on Good Clinical Practice (GCP) principles, and people, physicians and researchers willing to participate in clinical trials [20]. However, each country within Latin America has its own regulations (some with significant bureaucracy) and all lack research funds, thus making it challenging to maintain the infrastructure and capacity to conduct clinical trials [21], [22]. Unfortunately, without a constant flow of studies, following successful efficacy vaccine trials in the more distant and recent past, many sites were dismantled and lost qualified personnel, development capacity and site qualification [21]. With the COVID-19 pandemic spreading, and Latin America, in particular Peru, Brazil and Mexico, becoming hotspots, there was a rapid surge in demand for large scale COVID-19 vaccine trials for which the region and the world were not fully prepared. When the COVID-19 pandemic was declared (on 11 March 2020), nearly half (∼44 %) of the 7770 vaccine clinical trials registered on ClinicalTrials.gov were conducted in North America with only 6 % conducted in Latin America. By 31 July 2020, just before this initiative started, 131 COVID-19 vaccine trials were registered on ClinicalTrials.gov globally, with 20 (15 %) being conducted in Latin America [23]. Public health measures to contain the pandemic had an impact on site infrastructure, space needed to ensure social distancing and personal protective equipment demands. Staffing levels had to be rapidly expanded and personnel trained in the conduct of large vaccine trials for which recruitment in record time was crucial. Lockdowns and a high demand for health care professionals to ensure medical care aggravated the accessibility of qualified site personnel, a key factor for ensuring quality in clinical trials. Training on GCP, local regulations, clinical study protocols, and SOPs was essential as well as direct and frequent communication with the sites. Furthermore, specific equipment demands varied between trials and vaccine type, so ensuring continuous access to supplies during this global shortage crisis posed an added challenge. Some of the key factors that define exemplary trial sites include fast recruitment, maintenance of high GCP standards and quality, and multidisciplinary involvement in the process [24], [25]. In addition to general clinical trial requirements, COVID-19 vaccine trials have specific needs such as infrastructure and additional personnel to cope with a high and fast recruitment never experienced by any site before. The expectation was to recruit 800-–1000 subjects per site per month. Furthermore, most sites were more experienced in paediatric trials than in adult trials, but adults were the initial target group for the development of the COVID-19 vaccines. This high enrollment impacts on many aspects: the need for a high number of site personnel proficient in remote data capture, including electronic diary cards to capture events in real time; specific storage and preparation requirements as some vaccines needed reconstitution and some mRNA vaccines needed transportation and storage at −70 °C thus demanding very rigorous cold chain control and specific freezers; logistics and personnel for processing large sample volumes; clinical trials material and sample shipment logistics in times when many flight routes were cancelled; and reduction in lead time for defining the site, training staff; recruiting and vaccinating participants, and in particular capabilities and resources for a diligent safety follow up as the most advanced vaccines were based on novel technologies, such as mRNA or viral vectors, with a very limited safety record [26]. Conscious that operational readiness focusing on the conduct of efficacy trials was paramount for a quick and successful development of COVID-19 candidate vaccines, BMGF established the COVID-19 site readiness initiative to help trial sites in Latin America, Africa, and Asia prepare and enhance their capabilities for conducting large scale COVID-19 vaccine trials with a high enrolment strategy and as per local and international guidelines. The initiative funded 3 PDPs: Instituto D’OR de Pesquisa e Ensino (IDOR), Rio de Janeiro, Brazil; the Program for Appropriate Technology in Health (PATH), Seattle, USA; and the International Vaccine Institute (IVI), Seoul, Korea. PATH and IVI covered the site readiness in Africa and Asia. This paper discusses how quick and efficient this Latin America COVID-19 site readiness initiative, the principal investigator of the grant and lead author, and IDOR, Brazil, as the PDP, was at building site capacity for COVID-19 vaccine trials in this region.
A grant proposal (ID: INV-021464, total amount US$1,610,000) was approved by BMGF [27] to provide funds to IDOR, one of the PDPs, for preparing clinical trial sites to run large scale COVID-19 vaccine trials in Latin America. The implementation of this site readiness initiative was planned to last approximately 4 months (August 2020 to November 2020). The funding was intended for preparing or expanding trial sites in terms of infrastructure, equipment, staff, and training, for participating in COVID-19 vaccine trials, helping sites to expedite and ensuring quality of the vaccine trials.
The purpose of this initiative was to build/ improve and qualify clinical trial sites for conducting large scale Phase 3 COVID-19 vaccine trials mainly in adults within as early as 4 weeks after approvals for the most experienced sites. To select sites for building capacity to conduct large scale vaccine trials, a feasibility form was developed for site selection and sent out to various sites in Latin America. These sites were identified through a combination of avenues: PDṔs previous clinical trials experience with sites; through VacciNet [28], a network of investigators and clinical trial sites in Latin America; screening registries for vaccine publications; knowledge of trials already conducted in the region; and via contact with clinical research organizations (CROs). Main criteria for the selected sites to receive funding included experience of site and principle investigator in conducting clinical trials with focus on vaccine or infectious diseases, well trained staff including in GCP and safety reporting; existing SOPs; existing or potential to rapidly improve the infrastructure to accommodate the specific needs for COVID-19 trials, such as social distancing and separate space for work-up of potentially infected study participants; ethical review by Institutional Review Boards (IRBs) constituted and chartered to standard norms, regulatory approval timelines and logistics to allow a quick study start; capabilities to vaccinate 800–1.000 adult participants within a month, and access to real-time and reliable COVID-19 epidemiological data for their capture areas (selection criteria and flow chart in Fig. 1, Fig. 2). The principal investigator of the grant and the management team reviewed the completed feasibility forms, considering and weighting various factors for final site selection. Experience in vaccine clinical trial conduct, internationally accepted ethical and regulatory standards and oversight, as well as time required to be ready and include the first subject in a COVID-19 vaccine trial were the key selection factor above the other deliverables, such as infrastructure, personnel involved, budget requirements, epidemiological data, processes followed, and site retention strategy. The final site list was shared and agreed with the project funder, BMGF.
Due diligence was done for all selected sites to map current site capacity and experience, identify gaps and resource needs, check their budget requirements, and negotiate budget funding, after which contractual agreements were executed, and funding implemented. Funding included the purchase of site and laboratory equipment (such as fridges, freezers, and centrifuges) and administrative materials (such as desktops, laptops, broadband connection and Wi-Fi routers, secure servers and connection), the construction or renovation of physical space, the hiring of staff, the delivery of training sessions, the monitoring of epidemiological data, and the external site validation, and site security. To assess training needs, the selected sites were classified in 3 tiers based on staff experience in clinical and vaccine trials, site facilities and equipment, priority and speed to initiate COVID-19 vaccine trials. Depending on this classification, tailored training sessions were prepared and delivered to all sites in preparation for the conduct of COVID-19 vaccine trials. Training sessions varied in scope, subject depth, and length according to each tier’s needs. The content and training were developed and approved under the supervision of the University of Siena, Italy, which pioneered the first internationally recognized Master in Vaccinology and Pharmaceutical Clinical Development (https://ifgh.org/educational-programs/masters/master-invaccinology/) and has specific modules and materials for this training which include key subjects such as GCP, disease awareness, COVID-19 vaccine clinical trials fundamentals, vaccine clinical trials fundamentals recruitment strategies and adherence, data management – CRF and monitoring, finance, investigational product (IP) and laboratory management, safety reporting, shipment, regulatory reports, inspections, and SOP for basic studies. Trained site personnel received a certificate from the University of Siena. To continuously assess the appropriateness of selected sites while the pandemic evolved, the following activities were additionally performed: (i) collection of COVID-19 epidemiological data and follow-up on a continuous basis; (ii) identification and documentation of data sources by country; and (iii) periodic surveys of selected sites using the COVAX dashboard for sharing new information on the sites and COVID-19, with biweekly dashboard updates and posting on the public COVAX platform set up by the CEPI [29].
Validation activities started after completion of the implementation phase to ensure the quality of the execution, the site readiness and the appropriate use of funds. Prior to validation visits the sites were to send in documentations including CV and training records of the principal investigator and site staff, documentation of the ethical and regulatory processes, site set up for safety follow up, a grant utilization report and a progress report documenting the activities carried out to address the initially identified gaps. Validation was done face-to-face wherever possible and, if not, virtually for each site by qualified consulting agencies. The process included interviews with the principal investigator and key site staff as well as a physical or virtual facilities tour of each site for assessing the site’s infrastructure and equipment. Based on the document submission, interviews and facility tour, an assessment report was written for each site to document: if and how they used the grant to improve their capabilities, if any inaccuracies in grant utilization were identified, and the outcome of the assessment. Operational activities The management of the grant’s full budget and the general activities for the Latin American region were closely followed by the grant PI and her management team which was composed of one managing senior scientific director; one project manager; a CRO; an agency specialized in performing audits and inspections; and one operational manager. The Operational manager ensured that all the data needed for the COVAX dashboard was gathered in a timely manner and was accurate; this included epidemiology, IT, reports, graphics, and metrics data. The finance and legal departments were responsible for setting up of contracts and grant wires with all selected sites. The sites were classified into tiers and a tailored training agenda was developed per tier. Training agenda, content and materials were reviewed and approved by the grant PI and the University of Siena. Throughout the project, regular updates were shared with BMGF and the other PDPs to ensure an open communication.
Overall, 34 sites were contacted across 10 countries for completion of the feasibility form. Of those, 22 (65 %) sites across 7 countries were included in the site readiness initiative: 3 sites in Mexico, 2 in Guatemala, 1 in Honduras, 2 in Dominican Republic, 5 in Colombia, 6 in Brazil, and 3 in Peru. The list of selected sites is provided in (Table 1). Twelve (35 %) sites from 3 countries were excluded as they did not fulfil the selection criteria (Fig. 1, Fig. 2). Characteristics of included sites based on feasibility questionnaire Critical to the selection of these sites was that all of those sites or their investigators had previous experience with vaccine trials or with clinical trials for infectious diseases within the previous 5 years. All had core permanent staff experienced in clinical trials and trained in GCP, the majority of the sites (83 %) also had access to temporary staff to cope with an increased demand in trials. Senior site personnel had, based on their level of experience, a level of autonomy in executing the trials, with the support of experienced coinvestigators and site coordinators. Most of the selected sites also had had previous audits or inspections without any critical findings (89 %); All sites were supervised by independent IRBs prioritizing COVID-19 clinical trials and constituted according to international regulations; all sites had adequate processes in place and qualified staff to ensure safety follow up. The existing or upgraded infrastructure reflected the specific requirements of a COVID-19 trial. The regulatory authorities of the countries where the selected sites were based had special provision for expedited review of clinical trial applications. All sites had access to epidemiological data on the COVID-19 pandemic of the country and mostly also of the site region, and the countries where the sites are based had no restrictions on importing the investigational vaccine or exporting biological specimens. In all countries, experienced CROs were present. The selected sites had government support for conducting COVID-19 trials and had access to suitable participants to ensure the target recruitment of 800–1.000 participants within 1 month. Furthermore, most sites had server security (72 %) systems in place, available and tested shipment courier providers (94 %), and all had access to a network of laboratories. Funding and training of selected sites: Results from the implementation phase Following site selection, contracts were finalized and the agreed funds were provided. The classification in tiers was primarily bases on staff experience and training requirements: 8 sites in Tier 1 – Advanced (trained on disease, COVID-19 vaccine clinical trials and GCP), 6 sites in Tier 2 – Intermediate (trained on disease, vaccine clinical trials, COVID-19 vaccine clinical trials, GCP, and SOP fundamentals), and 8 sites in Tier 3 – Basic (trained on disease, deeper knowledge on the fundamentals of vaccine clinical trials, safety, surveillance, data management and operational aspects, COVID-19 vaccine clinical trials, GCP, and SOP fundamentals). Three of the 8 sites in Tier 1 had their training prioritized as they had previously been approached by sponsors to conduct COVID-19 vaccine trials and had to be ready within a shorter timeframe so as not to jeopardize their awards. A total of 629 staff were trained and certified, including key site staff (site investigators, sub-investigators, coordinators, study operational managers, nurses, laboratory and data management personnel, safety, quality, surveillance and call center teams). Staff who already had a current valid GCP certificate did not need to take the GCP exam but had to participate in GCP workshops, SOP fundamentals, disease awareness and COVID-19 vaccine fundamentals in vaccine clinical trials and all other trainings. If participants did not pass the GCP exam, they had to repeat it to get the certification. For continuity, the training material was shared with investigators at all sites to ensure training of any staff who could not attend delivered sessions and to train any new staff recruited for upcoming trials. Qualified trainers were identified for this activity. Grant funds were utilized for buying equipment, hiring human resources, and for building and renovating space, as per grant request agreement. See Fig. 3 for details on the usage of grant budget. Overall, the highest proportion of funds were used for building and renovating space (46 %) to meet the specific SARS-CoV-2 precautionary requirements and for buying equipment (36 %), and the remaining 18 % were used for human resources. The top categories reported were ‘medical consultation room’ under space and renovations; ultra-low temperature freezers (-80 °C), computers, power generators, freezers (-20 °C), and fridges (2 °C to 8 °C), under the equipment category; and biomedical/pharmacist/biologist, doctor, nurse, security, and site coordinator, under human resources. The individual grant dispensed to the sites ranged from US$35,000 to US$130,000.
A virtual tour was performed at all 22 sites and included checking of waiting rooms, consultation rooms, cold rooms, IP, management rooms, vaccination rooms, laboratories, machinery area (power and IT), offices and administrative spaces, designated archival area and satellite sites. Key recommendations following from the virtual tours were: the creation of site SOPs for cold chain maintenance and contingency plans (8 [36 %] sites), clinical trial material transfer to satellite sites (3 [14 %] sites), and sample transfer to main site/laboratory (2 [9 %] sites). Other recommendations for site process for improvement were related to other SOP areas: informed consent form (ICF) process, site staff training, equipment maintenance, and calibration, contingency planning in case of a disaster, and clinical trial management in case of a pandemic. At the end of this site readiness initiative project, 21 of 22 (95 %) sites were ready to conduct COVID-19 vaccine trials as per the requirements of this project. Importantly, as the COVAX website was regularly updated with site readiness details, by the end of the project each of these 21 sites already had agreements in place or were in discussions with sponsors to conduct large scale COVID-19 vaccine trials. The remaining site was a completely new site in Bogota, Colombia, that was created and is managed by CEIP, Investigational Center for Pediatric Infectious Diseases, based in Cali. The new CEIP site in Bogota was created solely for COVID trials using the budget received from this grant. This site was certified by the Colombian NRA in May 2021.
Since the COVID-19 pandemic was declared, an immense global collaborative effort has been made to advance science and stop, or at least control, its devastating effects on human health and health systems worldwide. The race to have an efficacious and safe vaccine available for a wider population, even if for emergency use only, was heightened. The high number of cases globally provided an important opportunity for efficacy trials to be conducted, but uncovered a lack of readiness to execute large studies in a short time in those areas which were most affected. This vital COVID-19 site readiness initiative, funded by BMGF, was set up to support and prepare clinical trials sites in HMICs and LMICs for the conduct of an unprecedented number of concurrent Phase 3 clinical trials with new COVID-19 specific infrastructure, physical capacity, and staffing requirements [29]. Considering its outcome, this initiative was highly successful. In total, 34 sites were mapped in 10 countries, and 22 (65 %) sites across 7 countries were selected to participate in this initiative. All 22 sites were ready to conduct large scale Phase 3 efficacy trials within 4 months of project start with one new site pending regulatory authority certification. This included 10 new independent investigational areas that were either the result of expansion of sites already located within the same building or independent satellite sites located in a different area from the main investigational site, created using this grant’s resources, and developed solely for the conduct of large vaccine trials. These totally new investigational sites/areas, now qualified for conducting large Phase 3 trials, are a legacy of the pandemic and are the result of a joint effort from BMGF, experienced research teams and local investigators that brings hope for development in time to help the pandemic. Despite the ongoing pandemic, all sites managed to get ready in both capacity building and infrastructure. Considering the number of COVID-19 vaccine trials ongoing in Latin America (Fig. 4) as of 2nd December 2021, we may assume that this initiative was a key contributor to the sharp rise in the numbers of such trials – a 4-fold increase from the time this initiative was kicked-off (Fig. 5 lists websites with information on COVID-19 studies). So far, few studies have been conducted on strengthening of clinical trial capacity through infrastructure upgrades, recruitment, and personnel training, thereby leading to improved ability for conducting future trials and thus, better health systems [30], [31], [32], [33], [34], [35]. The present study, conducted in Latin America, is a first of its kind to focus on a rapid improvement in the capacity of clinical trial sites to run large scale vaccine trials, particularly considering the current global burden of the COVID-19 pandemic and the intrinsic challenges when dealing with respiratory infectious diseases. To successfully achieve the goals of this project in record time, we focused on: 1) appropriate site selection, guided and supported by the feasibility questionnaire created; 2) assessment of training needs and gaps, with development and delivery of training at the level required to each selected site; 3) securing and providing the grant budget requested by the sites for purchase of equipment and consumables. Of note, it was essential to have an experienced team in place to plan, implement and deliver the training, and meet the main objective of the project – to ensure timely vaccine development. The assessment of training needs and gaps, and delivery of tailored training to each site were critical to accomplish the high and fast subject enrolment required for Phase 3 COVID-19 vaccine trials. Recruitment conditions during a pandemic are much more strenuous and consuming on site staff, and training is key to guarantee not only recruitment but also quality and accuracy of data. Conducting clinical trials at inadequate sites can lead to delays in starting trial related activities, under-recruitment, poor data quality. This can put study participants at risk, result in inefficiencies and wastage of time, cost, and resources, besides the high risk of invalidating the data collected and possibly even the whole trial and, in this case, delaying the hope for an authorized/licensed product to start controlling the pandemic. Thus, selecting a suitable clinical trial site and investigational team, getting it ready to run a clinical trial, and ensuring its performance are important steps to ensure the successful completion of a trial [36]. Further, investigator-dependent factors, such as previous experience, concurrent workload, and publications record, and ease of trial approval are critical factors that determine site selection [37]. Thus, a feasibility questionnaire was developed and sent to all Latin American mapped sites as a means of assessing their status, identifying gaps, and selecting sites that could create new areas or satellite sites, was crucial for expanding vaccine trials capacity in this region. All sites and/or investigators were selected based on the robust feasibility characteristics such as experience in the conduct of vaccine trials or other clinical trials in infectious disease areas in the last 5 years, access to laboratories, public health measures for COVID-19 as per regulatory requirements, no restrictions to import trial drug or to export biological specimens, and access to risk group. GCP deviations are commonly found in the review process of new submissions, highlighting the need for in-depth and intensive training while preparing for clinical trials [38], [39]. Thus, the present initiative also had a special focus on GCP training for the site staff involved in the conduct of these trials. The site selection process was based on experienced CRO qualification processes of a site but had important differentiations: possibility for early and repetitive interaction with authorities with respect to study approvals, epidemiological data access, ease of import and export is not a standard procedure for a CRO. Timelines between selection, upgrade, qualification, and study start were very much compressed compared to standard CRO metrics. The specific requirements for a COVID-19 treatment or vaccine trial were to be respected and implemented. A virtual inspection of the sites was innovative based on the travel restrictions and social distancing requirements. The training of the site staff was tailor made for the experience of the site which is not standard practice. And finally the costs for the validation of 22 sites across 7 countries were materially less than typical CRO charges. This is an original and unique initiative, in a time of no precedents in the history of clinical development, to prepare investigational sites to deliver support to bring vaccines to the population in record time, while the pandemic was ongoing. The site readiness project team, assembled in a short period of time by the lead author and grant PI, was crucial to achieve the goals of the project. This experienced team was balanced with different talents within clinical development to draw a very solid and thorough plan with strict timelines, implementation path, back-up solutions to respond to unexpected challenges and expected outcomes. The strategic plan was reflected in the deliverables and timelines - for each specific aspect, generating clusters of trainings and tailormade agendas to prepare and build sites with international standards for Phase 3 clinical development. All international and national guidelines with regards to GCP, ICH and local regulatory and ethical rules were respected. Thus, the newly created IDOR site readiness team for this project, together with the University of Siena, developed and tailored the training curricula according to the selected sites’ prior experience with vaccines/clinical trials, in line with the aim of preparing these sites for COVID-19 trials or vaccine trials in general. This also provided an academic supervision and certificates for those trained, thus enriching and improving their fundamental knowledge in this field of work. This type of specialized and tailor-made training is not generally universally available and therefore it was a key motivator for the sites and staff as well as for the local scientific communities and authorities in countries that supported this initiative either in-kind or actively. A decision to validate the sites by an independent organization was taken to ensure that quality with regards to personnel and site infrastructure were met to cope with the international standards and that data generated in those sites would immediately contribute to regulatory dossiers for COVID-19 vaccine registrations. Those sites were inserted in the COVAX dashboard where the developers of the COVID-19 vaccines were able to search for qualified sites. The site validation was one more warranty for sponsors that the site selected would be immediately effective and produce quality and regulatory approvable data. Therefore, sponsors could focus on the product, submissions and clinical development plans to bring forward new products to the general population. As validation outcomes, minor aspects on SOP recommendations were highlighted, which ensured the quality training and readiness of both site and personnel. No major or critical aspects were highlighted during the validation. At the time of validation, the results confirmed that all sites had utilized the greatest amount of their grant (81 %) for site infrastructure improvement (equipment, and space and renovation), and the rest on human resources, to increase site enrolment capacity. Part of the grant was reserved for hiring personnel once a study was due to start at the site or to pay the salaries of site personnel while negotiations with vaccine trial sponsors were still ongoing, to retain the trained personnel. Several challenges were encountered during the conduct of this initiative, none of which caused a serious hindrance or delay to the project. Some of the countries initially mapped as potential participants had to be dropped due to lack of qualified personnel or because it would be too costly or time consuming to create the infrastructure required. Moreover, due to the ongoing pandemic and the need to accelerate vaccine development, selected sites were competing not only with healthcare systems but also with other sites, sponsors and CROs already running COVID-19 studies for qualified and experienced personnel. Attention had to be given to finding and retaining site personnel. Considering the worldwide shortage of equipment and consumables, sites reported the same difficulties: freezers, centrifuges, laptops, PPE and other equipment that were either sold out or priced 4-10x the normal market value. Investigators and site personnel had to use their negotiating skills to get their goods in time and within granted budget; direct contact with distributors proved important to get them to commit to supplying the required goods. The commitment, dedication, and enthusiasm of the investigators were crucial for success – even though they had no guarantee that they would be awarded the conduct of a COVID-19 vaccine trial. Some of the limitations of this site readiness initiative were due to the restrictions imposed in this pandemic situation: the majority of the interactions with the sites were remote (online), particularly the feasibility assessments and validation activities; most sites had predominantly experience in paediatric vaccine trials and less so in adults; the urgency to have the sites ready to run clinical trials within a few weeks to months due to the high demand for COVID-19 trials; and the lack of a blueprint for this particular type of site readiness initiative in such a particular situation in which healthcare systems were already overloaded.
To overcome the COVID-19 pandemic, fast vaccine development is crucial as a cornerstone in pandemic management. To develop a safe and efficacious vaccine, it must be tested through randomized controlled clinical trials with a large sample size. This site readiness initiative was carried out to build or improve clinical research sites to expand the capacity to conduct high-quality, large scale COVID-19 efficacy vaccine trials in Latin America, which had been a major contributor in these trials in the past. Through this initiative, suitable sites were selected in the Latin American region according to the feasibility criteria; gaps, mainly related to infrastructure, training and human resources, were identified, after which funding and training were implemented. Validation was performed to ensure those initially identified gaps were resolved. A total of 22 sites, including 10 new investigational sites, or areas within sites, were capacitated, validated and qualified within 4 months. The project was highly successful: 21 of the 22 (95 %) sites are currently involved in vaccine efficacy trials. In addition, this initiative also provided important information related to current barriers and their resolutions to enable the building of COVID-19 clinical trial sites in the Latin America region and worldwide. Hopefully, mechanisms will be put into place in order to keep those sites as reference of excellence in vaccine trials through continuous involvement in vaccine trials beyond COVID-19. One of the longer term effects of this grant project is further institutional strengthening which hopefully will be sustainable. Clinical Trials are the primary way to generate actionable evidence for healthcare interventions. The COVID-19 response, led by initiatives such as BMGF site readiness, has demonstrated the critical importance of clinical trials, highlighting the need for continuous support for an international framework on capacity building in clinical development. International mechanisms for collaboration and coordination of clinical trials network must be strengthened. The sites that are part of this initiative need continuous investment to keep enhancing clinical trial capability, specifically in HMICs and LMICs, to ensure capacity with quality is available where it is most needed to better address ongoing global health issues and to respond rapidly to possible health threats. This successful initiative will remain as one of the legacies of the COVID-19 pandemic. We hope that it can also serve as a blueprint on how to find, select, qualify and validate sites for clinical trials of different complexities in public health emergency situations. Funding sources This work was supported by the Bill and Melinda Gates Foundation, Seattle, Washington, USA (Grant No INV-021464). Data Sharing Statement The Bill and Melinda Gates Foundation group has regularly shared the project data via the COVAX website - Dashboard of COVAX-Supported Sites Ready for Clinical Trials (https://epi.tghn.org/covax-overview/clinical-science/clinical/#ref).
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: [Sue Ann Costa Clemens is a Professor of Pediatrics Infectious Diseases, Head of the Institute for Global Health University of Siena and Director Vaccine Group Oxford-Brazil, Visiting Professor in Global Health at Oxford University, Department of Pediatrics and Senior Adviser and consultant at the Bill and Melinda Gates Foundation. Ana Keiko Sekine is a Director of Clinical Operations – COVID Vaccine Program, Moderna Inc. as of August 2021. Fernanda Tovar-Moll is the CEO and a member of the Board of Trustees of Instituto D’Or de Pesquisa e Ensino. Ralf Clemens is an adviser to BMGF and member of the Board of Trustees of IVI.]. |
|
PMC9647653 | Yaqin Zhang,Xiangyu Quan,Yingchun Li,Hangyu Guo,Fange Kong,Jiahui Lu,Lirong Teng,Jiasi Wang,Di Wang | Visual detection of SARS-CoV-2 with a CRISPR/Cas12b-based platform | 10-11-2022 | SARS-CoV-2,CRISPR/Cas12b,Point-of-care,High-throughput test,COVID-19, Corona Virus Disease 2019,SARS-CoV-2, severe acute respiratory syndrome coronavirus 2,RT-PCR, reverse transcription-polymerase chain reaction,POC, point of care,LAMP, loop-mediated isothermal amplification,RPA, recombinase polymerase amplification,RCA, rolling circle amplification,CRISPR, clustered regularly interspaced short palindromic repeats,Cas, CRISPR-associated,AuNP, Gold nanoparticle,TCEP, Tris(2-carboxyethyl) phosphine,DMEM, Dulbecco's modified Eagle medium,TEM, transmission electron microscope,DEPC, Diethypyrocarbonate | The coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused a global pandemic, highlighting the unprecedented demand for rapid and portable diagnostic methods. Clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated (Cas) proteins-based platforms have been used for the detection of pathogens. However, in further applications and research, due to multiple steps needed, many methods showed an increased risk of cross-reactivity. The thermostable Cas12b enables the combination of isothermal amplification and CRISPR-mediated detection, which could decrease the risk of cross-contamination. In this study, we developed a portable and specific diagnostic method that combined the gold nanoparticle (AuNP) with thermal stable CRISPR/Cas12b-enhanced reverse transcription loop-mediated isothermal amplification (RT-LAMP), which is called SCAN, to distinguish the N gene of SARS-CoV-2 from flu gene. We validated our method using RNA from cells transfected by plasmids. We could easily distinguish the positive results by the naked eye based on the strong molar absorption coefficient of AuNP. Moreover, SCAN has the potential for high-throughput tests owing to its convenient operation. In sum, SCAN has broken the site and equipment restrictions of traditional detection methods and could be applied outside of hospitals and clinical laboratories, greatly expanding the test of COVID-19. | Visual detection of SARS-CoV-2 with a CRISPR/Cas12b-based platform
The coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused a global pandemic, highlighting the unprecedented demand for rapid and portable diagnostic methods. Clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated (Cas) proteins-based platforms have been used for the detection of pathogens. However, in further applications and research, due to multiple steps needed, many methods showed an increased risk of cross-reactivity. The thermostable Cas12b enables the combination of isothermal amplification and CRISPR-mediated detection, which could decrease the risk of cross-contamination. In this study, we developed a portable and specific diagnostic method that combined the gold nanoparticle (AuNP) with thermal stable CRISPR/Cas12b-enhanced reverse transcription loop-mediated isothermal amplification (RT-LAMP), which is called SCAN, to distinguish the N gene of SARS-CoV-2 from flu gene. We validated our method using RNA from cells transfected by plasmids. We could easily distinguish the positive results by the naked eye based on the strong molar absorption coefficient of AuNP. Moreover, SCAN has the potential for high-throughput tests owing to its convenient operation. In sum, SCAN has broken the site and equipment restrictions of traditional detection methods and could be applied outside of hospitals and clinical laboratories, greatly expanding the test of COVID-19.
Over the past two years, coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has emerged as a global pandemic [1]. As of February 20, 2022, there have been over 422 million SARS-CoV-2-confirmed cases worldwide with over 5 million deaths [2]. Although the vaccines and therapeutic agents have been developed, early detection of SARS-CoV-2 is still critical to prevent the infection and control the pandemic. Of note, real-time reverse transcription-polymerase chain reaction (RT-PCR) is still the gold standard of nucleic acid diagnosis due to its sensitivity and specificity [3]. However, RT-PCR needs trained technicians and a thermal cycling controller, which has limited its onsite application. Nowadays, the SARS-CoV-2 rapid antigen test has been applied widely all around the world, which is fast and convenient while with low sensitivity and specificity (106 copies/mL, 77.4–96.8% specificity) [4]. Therefore, there is still an unprecedented need to develop detection methods that are rapid, convenient, sensitive, and portable to use at the point-of-care (POC). As an alternative to the thermal cycling amplification, a plethora of isothermal amplification methods, such as recombinase polymerase amplification (RPA) [5], loop-mediated isothermal amplification (LAMP) [6,7], and rolling circle amplification (RCA) [8], have been applied to detect the SARS-CoV-2 which could get rid of the need of expensive equipment. Recently, clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated (Cas) protein has been applied to molecular diagnostics. These endonucleases guided by crRNA have the target-dependent cis- and trans-cleavage activities, resulting in a high-turnover signal amplification mechanism. Several Cas-based assays have been established, such as Cas12a-based DNA endonuclease-targeted CRISPR trans reporter (DETECTR) [9], Cas13a-based specific high-sensitivity enzymatic reporter unlocking (SHERLOCK) [10], Cas12a-based 1-h low-cost multipurpose highly efficient system (HOLMES) [11]. These Cas-based methods were usually divided into two steps, target nucleic acid amplification and CRISPR recognition, which might bring a risk of cross-reactivity and false positive. Hereupon, numerous one-pot detection methods have been developed to cope with the inaccuracy caused by multistep reactions; Zhang's group developed STOPCovid.v2 to detect SARS-CoV-2 with lateral-flow readout by combining RT-LAMP and CRISPR/Cas12 b in one-pot [12]; J.S. Park et al. developed deCOViD to detect SARS-CoV-2 based on fluorescence which combined RT-RPA and CRISPR/Cas12a [13]; W. Feng et al. developed an integrated RT-RPA/CRISPR/Cas12a assay to detect SARS-CoV-2 with end-point visualization and fluorescence [14]. Most of these methods use fluorescence and microfluidic devices to display the results, which limit the wide application at POC. The Cas12b which came from Alicyclobacillus acidiphilus has been reported to be more thermostable [15]. Compared with the previous methods based on Cas12a [16], the reaction temperature (31 °C–59 °C) of Cas12b protein is compatible with RT-LAMP to initiate the reaction in a homogeneous reaction system. This property could make the RT-LAMP and CRISPR/Cas12b detection be performed isothermally without sophisticated instruments and professional operators. Furthermore, the simplified operation process would make it possible to be an automatic platform. Here, we developed a visual assay for the detection of SARS-CoV-2 N gene RNA by combining the gold nanoparticle (AuNP) with thermal stable CRISPR/Cas12b-enhanced RT-LAMP (SCAN). The visual detection was based on the strong molar absorption coefficient of AuNP [17]. AuNPs were pre-assembled with two different sulfhydryl DNA (DNA1 and DNA2). Since the sequence of linker-ssDNA was complementary to partial sequences of DNA1 and DNA2 at 5′ and 3’ ends respectively, the distance between the two AuNP probes was shortened resulting in a color change (Fig. 1 ). When the sample contained target RNA sequences, SARS-CoV-2 N gene RNA was reverse transcribed and amplified by RT-LAMP. The amplification product was then recognized by Cas12b/crRNA which could activate the trans-cleavage activity of Cas12b to cut linker-ssDNA non-specifically. Therefore, the AuNP probes remained mono-dispersed and red. In the absence of the target nucleic acid, the linker-ssDNA would not be cleaved by Cas12b, and the AuNP-DNA1/2 would be cross-linked by linker-ssDNA resulting in a color change from red to purple.
AuNP was purchased from Suzhou Tanfeng Graphene Technology Co., Ltd. (Suzhou, Jiangsu, China). The crRNA (Table S1) was ordered from Bio-Lifesci (Guangzhou, Guangdong, China). Other nucleic acid (Table S1) was ordered from Sangon Biotech (Shanghai, China). NaCl was purchased from Sinopharm Chemical Reagent Co., Ltd. (Beijing, China). Na2HPO4, NaH2PO4 and KCl were purchased from Beijing Chemical Works (Beijing, China). WarmStart® LAMP Kit (DNA & RNA), Isothermal Amplification Buffer Pack, WarmStart® RTx Reverse Transcriptase, Bst 2.0 WarmStart® DNA Polymerase, and MgSO4 were obtained from New England Biolabs (Beijing, China). Tris (2-carboxyethyl) phosphine (TCEP) was obtained from Aladdin (Shanghai, China). AapCas12b was obtained from Magigen Biotechnology Co., Ltd. (Guangzhou, Guangdong, China). BM2000+ DNA Marker was obtained from Biomed (Beijing, China). Diethypyrocarbonate (DEPC)-treated water (DNase/RNase free), Tris-HCl, (NH4)2SO4 and Tween-20 were obtained from Beyotime (Shanghai, China). Dulbecco's modified Eagle medium and fetal bovine were obtained from Procell Life Science&Technology Co., Ltd. (Wuhan, Hubei, China). Tris-EDTA (TE) Buffer, penicillin/streptomycin and GeneJET RNA Purification Kit were acquired from Thermo Fisher Scientific (Shanghai, China). COVID-19 RNA reference materials (high concentration) were obtained from the National Institute of Metrology of China (Beijing, China).
Transmission electron microscope (HT7800, Hitachi, Japan); centrifugal machine (MiniSpin®, Eppendorf, Germany); qPCR machine (CFX 96™, Bio-rad, America); Micro UV spectrophotometer (NanoDrop one, Thermo Fisher, China); gel Imaging System (GelDoc Go, Bio-Rad, America); microplate reader (Epoch 2, Bio-Tek, America); metal bath (HB120–S, DLAB Scientific Co., Ltd., China); well plate incubator (HCM100-Pro, DLAB Scientific Co., Ltd., China).
The sulfhydryl DNA1 and DNA2 were dissolved in TE Buffer to 100 μM, and the DNA was reduced by TCEP at the ratio of 1:100 (DNA: TCEP) for 30 min at room temperature. The combination of AuNPs and sulfhydryl DNA followed the salt aging method according to previously reported protocols [18]. First, AuNPs were incubated with the sulfhydryl DNA in the ratio of 1:200 (AuNPs: DNA) overnight at room temperature. Thereafter, the mixture was incubated with 2 M sodium phosphate buffer (2 M of NaCl, 200 mM of Na2HPO4 and NaH2PO4, pH 7.4) to reach a final concentration of 200 mM NaCl (the whole process was divided into four steps, and every interval was no less than 1 h). The salt aging process was allowed to incubate overnight at room temperature. At last, the unbound DNA was removed by centrifugation (12,000 rpm, 20 min, three times) using 0.01 M Tris-HCl buffer (pH 7.4). The sulfhydryl DNA modified AuNPs conjugate was resuspended in 0.01 M sodium phosphate buffer (pH 7.4) and stored at 4 °C.
The sequence of linker-ssDNA was designed to be partially complementary to the sequences of AuNP-DNA1 and AuNP-DNA2 at 5′ and 3’ ends respectively. To investigate whether the linker-ssDNA could cross-link AuNP-DNA1 and AuNP-DNA2, linker-ssDNA was added to the preassembled AuNP-DNA1 and AuNP-DNA2 mixture to reach the ratio of 50: 1: 1 (linker-ssDNA: AuNP-DNA1: AuNP-DNA2). The 5 M NaCl was added to the mixture to reach the concentration of 200 mM of NaCl to accelerate the speed of the cross-link of linker-ssDNA and AuNP-DNA1/2. The color change of the mixture was observed after incubating at room temperature for 5 min and transmission electron microscope (TEM) images were taken to confirm whether the linker-ssDNA could cross-link AuNP-DNA1/2.
To optimize the Cas12b assay, the dsDNA was used to verify the feasibility of the Cas12b/crRNA detection. Firstly. Cas12b was preassembled with crRNA in 1 × buffer (20 mM Tris-HCl, 10 mM (NH4)2SO4, 50 mM KCl, 2 mM MgSO4, 0.1% Tween-20, pH 8.8) for 10 min at 37 °C. The assays were performed with serial dilutions of dsDNA, 400 nM reporter DNA, 1 × buffer, 31.25 nM AapCas12b, and 31.25 nM crRNA in a total volume of 25 μL. The real-time fluorescence was detected by a qPCR machine every 1.5 min at 60 °C. The optimization of the reaction buffer and temperature was conducted by real-time fluorescence detection. For KCl, Cas12b was preassembled with crRNA in 1 × buffer (20 mM Tris-HCl, 10 mM (NH4)2SO4, 2 mM MgSO4, 0.1% Tween-20, pH 8.8) with different concentrations of KCl (0, 25, 50, 100 mM) for 10 min at 37 °C. For MgSO4, Cas12b was preassembled with crRNA in 1 × buffer (20 mM Tris-HCl, 10 mM (NH4)2SO4, 50 mM KCl, 0.1% Tween-20, pH 8.8) with different concentrations of MgSO4 (0, 5, 10, 20 mM) for 10 min at 37 °C. The reaction was carried out as described above. For temperature, the reaction was carried out at different temperatures (50 °C, 55 °C, 60 °C, and 65 °C).
293 T cells (a human renal epithelial cell line, obtained from the American Type Culture Collection, No CRL-11268) was cultured with Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum and 1% penicillin/streptomycin in a complete humidity incubator with CO2/air (5%/95%) at 37 °C. 293 T cells were seeded into 6-well plates at the density of 3 × 105 cells/well and then transfected by a plasmid with the N gene of SARS-CoV-2 at a dose of 2 μg per well, the negative group was set to be transfected by the plasmid without the N gene of SARS-CoV-2 at the same dose. The RNA was then extracted using an RNA Purification Kit according to the manufacturer's instructions and the concentration was measured by a micro UV spectrophotometer.
The amplification of SARS-CoV-2 RNA was performed with the WarmStart® LAMP Kit (DNA & RNA) according to the manufacturer's instructions. Briefly, the reaction, contained 12.5 μL of 2 × WarmStart® LAMP Master Mix reaction premix, 2.5 μL of 10 × primers mixture, 0.5 μL of 50 × fluorescent dye, and 1 μL of target RNA (different concentrations of COVID-19 RNA reference material and the RNA from transfected cells), 8.5 μL of water, was performed using a real-time PCR machine at 65 °C. Then, 2.5 μL of the amplification product was used to activate the Cas12b for fluorescent or visual assay. The fluorescence assay was performed as described above. For visual detection, the reporter DNA was replaced by linker-ssDNA, and the premixed AuNP-DNA1/2 solution was added to the reaction system at the ratio of 50: 1: 1 (linker-ssDNA: AuNP-DNA1: AuNP-DNA2) when the Cas12b/crRNA detection finished (30 min, 60 °C).
The RT-LAMP amplification products were then tested and verified using 2% agarose gel electrophoresis. Images were photographed by a gel Imaging System.
The SCAN reaction master mix consisted of the following components: 1 × Isothermal Amplification Buffer (20 mM Tris-HCl, 10 mM (NH4)2SO4, 50 mM KCl, 2 mM MgSO4, 0.1% Tween-20, pH 8.8), 1.4 mM dNTPs, 8 units of Bst 2.0 WarmStart® DNA Polymerase, 7.5 units of WarmStart® RTx Reverse Transcriptase, 100 nM Cas12b protein, 100 nM crRNA, 400 nM fluorescent reporter, 0.8 μM FIP/BIP primers, 0.2 μM F3/B3 primers, 0.2 μM LoopF/B primers and 8 mM MgSO4. The detection was performed at 60 °C in a qPCR machine with fluorescent measurements every 1.5 min. For visual detection, the whole reaction was performed at 60 °C on a metal bath for 2 h, then the premixed AuNP-DNA1/2 solution was added to the reaction system as described above, and the phenomenon was recorded by a mobile phone.
A 96-well plate was used to detect the N gene of SARS-CoV-2. For each component reaction of SCAN, the 96-well plate was then incubated in a well plate incubator at 65 °C for 30 min (RT-LAMP reaction) and 60 °C for 30 min (CRISPR/Cas12b detection). For the SCAN assay, the 96-well plate was incubated in a well plate incubator at 60 °C for 120 min. Then the premixed AuNP-DNA1/2 solution was added to the 96-well plate after reaction in the visual detection group. The color change was observed and the absorbance of the mixture was detected at 520 nm and 560 nm using a microplate reader.
To demonstrate the utility of the AuNPs-based visual detection, the AuNPs were modified by two different sulfhydryl DNAs (v-DNA1 and v-DNA2), which could be cross-linked by a linker-ssDNA. The solution of AuNP-DNA1/2 was red in the absence of linker-ssDNA, and the color changed from red to purple in 5 min when the linker-ssDNA was mixed with pre-assembled AuNP-DNA1/2 due to the shortened distance between AuNPs (Supplementary Fig. 1A). The photograph of TEM has confirmed that the color change was caused by agglomerated AuNPs in the presence of linker-ssDNA (Supplementary Fig. 1B).
To reduce the risk of cross-reactivity and increase the possibility of application at POC, the SCAN assay was designed to combine the RT-LAMP with CRISPR-mediated detection. It is necessary to optimize the reaction conditions which could satisfy the need for each reaction (Fig. 2 A). The trans-cleavage activity of CRISPR/Cas12b was evaluated by a fluorescent assay using the synthetic dsDNA as the target. As shown in Fig. 2B, Cas12b could be activated by target dsDNA at concentrations of at least 1 nM and 10 nM dsDNA was chosen for the following optimization. To acquire stable and efficient reaction conditions, we next optimized the reaction buffer in which RT-LAMP and CRISPR-mediated detection could occur simultaneously. The isothermal amplification buffer of RT-LAMP was used as a basis for the optimization because it has a similar composition to the buffer for CRISPR/Cas12b detection. It was shown that 50 mM KCl was the most efficient concentration for the Cas12b assay system (Fig. 2C). For Mg2+, the best signal quality was obtained at 10 mM and 20 mM (Figs. 2D), and 10 mM MgSO4 was chosen as the final reaction concentration considering costs. Since the most suitable temperature for Cas12b to maintain the highest trans-cleavage activity is from 31 to 59 °C according to the previous report [15], and the optimum reaction temperature for RT-LAMP is 60–65 °C, we then investigated the temperature of the CRISPR/Cas12b assay from 50 to 65 °C with an interval of 5 °C. Notably, the best trans-cleavage activity of Cas12b could be observed at 55 °C (Fig. 2E). In addition, the Cas12b at 50 and 60 °C also showed strongly trans-cleavage activity, but dropped significantly at 65 °C. Considering the optimum temperature of RT-LAMP, we selected 60 °C as the final temperature for the further experiment.
The diluted COVID-19 RNA reference materials were used to validate the SCAN assay. We performed the assay step by step, containing RT-LAMP, CRISPR/Cas12b activation, and visual detection. First, flu virus RNA was used as the negative control to demonstrate the feasibility of RT-LAMP. The amplification curve of serially diluted SARS-CoV-2 N gene RNA from 4 × 103 copies/μL to 4 × 100 copies/μL rose sharply and plateaued in 20 min (Fig. 3 A), and the amplification products were further confirmed by gel electrophoresis (Supplementary Fig. 2), indicating the SARS-CoV-2 RNA could be successfully amplified by RT-LAMP. We next tested if the amplification products could activate the trans-cleavage activity of Cas12b. As shown in Fig. 3B, the increased fluorescence could be observed in the presence of the amplification products, indicating the amplification products were specific. Finally, a difference in visual signal could be noticed between SARS-CoV-2 and flu virus (Fig. 3C). The multistep SCAN has also been performed in a 96-well plate with a temperature control device to verify the potential for high-throughput applications. As shown in Fig. 3D, the positive and negative samples were successfully discriminated by color change. The NTC and negative group turned to light purple color and the positive group (4 × 100–4 × 103 copies/μL) maintained pink color. The ratio of A520/A560 at 4 × 100 copies/μL of target RNA was consistent with the visual results above (Fig. 3E). Collectively, these results indicated that the target RNA could be successfully detected by multistep SCAN.
Since all component reactions in the SCAN assay have been verified and the reaction conditions have been optimized, we combined the RT-LAMP and CRISPR/Cas12b detection in one tube to simplify the workflow. First, the sensitivity and reliability of the one-step SCAN assay were evaluated. The results showed that 40 copies/μL of COVID-19 RNA reference material could be detected in 90 min by fluorescent assay (Fig. 4 A). In contrast, 400 copies/μL could be distinguished against the negative control with a visual assay (Fig. 4B). We also evaluated the sensitivity using a well plate incubator (Fig. 4C). The 96-well plate results correlated well with those observed by the naked eye in tubes, briefly, 400, 1000, and 4000 copies/μL RNA showed an obviously pink color and the other groups turned to purple color (Fig. 4D and E). Collectively, these results showed that the RNA of SARS-CoV-2 could be detected specifically using the one-step SCAN assay.
We have confirmed that the COVID-19 RNA reference material could be successfully detected by the SCAN assay. Then we evaluated the clinical feasibility of the SCAN assay by using the cells transfected by a plasmid with the N gene of SARS-CoV-2 to simulate the clinical samples. First, the RNA was detected by a multistep SCAN assay that we performed RT-LAMP, CRISPR/Cas12b detection, and visual detection separately to confirm the feasibility (Supplementary Fig. 3A). Consisted with the fluorescence results (Supplementary Fig. 3B), the presence of purple was observed by the naked eye in tubes (Supplementary Fig. 3C), indicating that the RNA from cells could be detected using a multistep SCAN assay. The color change in the 96-well plate (Supplementary Fig. 3D) and the increase of A520/A560 (Supplementary Fig. 3E) also suggested no cross-reactivity with the negative control group. Next, we performed the one-step SCAN assay directly to detect SARS-CoV-2 RNA extracted from transfected cells, which were serially diluted. The fluorescent results (Fig. 5 A) and visual detection results showed that our assay could distinguish positive samples when the concentration of RNA was more than 1 ng/μL (Fig. 5B). Finally, to demonstrate the feasibility of the assay for high-throughput assay and clinical application, we employed the assay to detect the RNA samples in the well plate. We observed a notable difference in color from pink to purple and the ratio of A520 to A560 between the RNA concentration of more than 1 ng/μL and other groups (Fig. 5C, Supplementary Fig. 4). Since the high-throughput reaction could be performed using a metal bath or a well plate incubator, it suggested the potential POC application.
In this study, we have developed a SCAN assay to detect SARS-CoV-2 in both visual-based and fluorescence-based platforms. The RT-LAMP and Cas12b detection are combined in one step owing to the heat-resistant activity of Cas12b. SCAN assay can satisfy several requirements of POC diagnosis. First, the procedure of SCAN is simple, that the RT-LAMP and CRISPR/Cas12b detection were performed in one tube to simplify the whole operation. Second, the SCAN assay is portable with inexpensive equipment, and the whole assay only relies on a metal bath. Furthermore, the results of the SCAN can be easily distinguished by naked eye due to the strong molar absorption coefficient of AuNP. Most important of all, the SCAN assay has a high-throughput potential to detect enormous amounts of virus samples simultaneously since the assay is isothermal and easy to be performed. To test the performance of SCAN in clinical settings, we used the extracted total RNA from cells transfected by SARS-CoV-2 to simulate the real extracted RNA, which showed good specificity and sensitivity. Given many RNA extraction-free methods have been developed to combine with RT-LAMP to detect SARS-CoV-2 [19,20], we expect the SCAN assay could be useful for diagnosis of COVID-19 at POC tests. In short, this method which is based on CRISPR/Cas12b and AuNP has the potential to assist in the rapid diagnosis and screening of patients with COVID-19.
Yaqin Zhang: Investigation, Methodology, Visualization, Validation, Formal analysis, Writing - Original Draft. Xiangyu Quan: Methodology, Validation, Investigation. Yingchun Li: Methodology, Validation, Investigation. Hangyu Guo: Validation, Investigation. Fange Kong: Investigation, Writing - Review & Editing. Jiahui Lu: Validation, Data Curation. Lirong Teng: Methodology, Visualization. Jiasi Wang: Conceptualization, Visualization, Supervision, Writing - Review & Editing. Di Wang: Conceptualization, Visualization, Writing - Review & Editing, Funding acquisition.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. |
|
PMC9647697 | Huan Zheng,Yang Dong,Huilan Nong,Liyuan Huang,Jing Liu,Xin Yu,Yaguan Zhang,Lina Yang,Ben Hong,Wu Wang,Jianmin Tao | VvSUN may act in the auxin pathway to regulate fruit shape in grape | 06-09-2022 | Abstract Fruit shape is an essential agronomic feature in many crops. We identified and functionally characterized an auxin pathway-related gene, VvSUN. VvSUN, which belongs to the SUN/IQ67-DOMAIN (IQD) family, localizes to the plasma membrane and chloroplast and may be involved in controlling fruit shape through auxin. It is highly expressed in the ovary, and the expression level 1 week before the anthesis stage is positively correlated with the fruit shape index. Functional analyses illustrated that VvSUN gene overexpression in tomato and tobacco plants changed fruit/pod shape. The VvSUN promoter directly bound to VvARF6 in yeast and activated ß-glucuronidase (GUS) activity by indole-3-acetic acid (IAA) treatments in grapevine leaves, indicating that VvSUN functions are in coordination with auxin. Further analysis of 35S::VvSUN transgenic tomato ovaries showed that the fruit shape changes caused by VvSUN were predominantly caused by variations in cell number in longitudinal directions by regulating endogenous auxin levels via polar transport and/or auxin signal transduction process variations. Moreover, enrichment of the 35S::VvSUN transgenic tomato differentially expressed genes was found in a variety of biological processes, including primary metabolic process, transmembrane transport, calcium ion binding, cytoskeletal protein binding, tubulin binding, and microtubule-based movement. Using weighted gene co-expression network analysis (WGCNA), we confirmed that this plant hormone signal transduction may play a crucial role in controlling fruit shape. As a consequence, it is possible that VvSUN acts as a hub gene, altering cellular auxin levels and the plant hormone signal transduction pathway, which plays a role in cell division patterns, leading to anisotropic growth of the ovary and, ultimately, an elongated fruit shape. | VvSUN may act in the auxin pathway to regulate fruit shape in grape
Fruit shape is an essential agronomic feature in many crops. We identified and functionally characterized an auxin pathway-related gene, VvSUN. VvSUN, which belongs to the SUN/IQ67-DOMAIN (IQD) family, localizes to the plasma membrane and chloroplast and may be involved in controlling fruit shape through auxin. It is highly expressed in the ovary, and the expression level 1 week before the anthesis stage is positively correlated with the fruit shape index. Functional analyses illustrated that VvSUN gene overexpression in tomato and tobacco plants changed fruit/pod shape. The VvSUN promoter directly bound to VvARF6 in yeast and activated ß-glucuronidase (GUS) activity by indole-3-acetic acid (IAA) treatments in grapevine leaves, indicating that VvSUN functions are in coordination with auxin. Further analysis of 35S::VvSUN transgenic tomato ovaries showed that the fruit shape changes caused by VvSUN were predominantly caused by variations in cell number in longitudinal directions by regulating endogenous auxin levels via polar transport and/or auxin signal transduction process variations. Moreover, enrichment of the 35S::VvSUN transgenic tomato differentially expressed genes was found in a variety of biological processes, including primary metabolic process, transmembrane transport, calcium ion binding, cytoskeletal protein binding, tubulin binding, and microtubule-based movement. Using weighted gene co-expression network analysis (WGCNA), we confirmed that this plant hormone signal transduction may play a crucial role in controlling fruit shape. As a consequence, it is possible that VvSUN acts as a hub gene, altering cellular auxin levels and the plant hormone signal transduction pathway, which plays a role in cell division patterns, leading to anisotropic growth of the ovary and, ultimately, an elongated fruit shape.
Fruits are the most valuable produce of horticultural crops. The size and shape of the fruit are crucial selection features in the course of developing new cultivars in the breeding process [1]. Wild fruits are usually small and round. Cultivars with varying fruit shapes and sizes have emerged as a result of gradual selective breeding and domestication [2]. Inheritance studies reveal that these traits are quite complex and are determined by multiple loci [3]. Researchers have undertaken comprehensive investigations on fruit shape and size as a vital criterion in the breeding of new cultivars to fulfill particular market demands, and a number of advancements have been made as a result of their efforts [4]. In recent decades, we have witnessed the cloning of several major quantitative trait loci (QTLs) related to fruit/grain size or shape in tomato [4, 5], papaya [5], cucumber [6], melon [7], peach [8], watermelon [9], cucurbits [10], rice [11, 12], and so on. QTLs identified in tomato are perhaps the best characterized for any fruit species; the ovate and sun loci influence elongated shapes, whereas the locule number (lc) and fasciated (fas) loci both modify locule number, and both influence the shape [13]. Of these QTLs, sun was identified as the primary locus influencing the elongated shape of the tomato fruit, explaining up to 58% of the phenotypic variation. As speculated by Xiao et al. [14], the origin of the locus was a consequence of a unique 24.7-kb gene duplication activity facilitated by the long terminal repeat retrotransposon rider. Fine mapping and cloning indicate the SUN gene as an affiliate of the IQ67 domain-containing family [14]. The plant-specific SUN/IQ67-DOMAIN (IQD) family has been identified as modulating the shape of fruits/grains among a variety of plant species. Fine mapping of a large F2 population of cucumber led to the identification of a putative gene, CsSUN, which is a homologous SUN gene for tomato fruit shape. Gene expression analysis has indicated that the long fruit expresses much more CsSUN as opposed to the round fruit [15, 16]. CmSUN-14, a homologous gene of CsSUN, could play a role in the development of melon fruit shape [16]. In watermelon, a 159-bp deletion mutation in the ClFS1 gene, which encodes the IQD protein, is crucial for determining the shape of the fruits [9]. OsIQD14 has been identified as a critical component in modulating microtubule reconfigurations in rice hull cells and, consequently, grain shape [12]. These findings suggested that the SUN/IQD-induced fruit shape may be modulated via a conserved mechanism [12]. Overexpression of SUN in tomato resulted in highly elongated parthenocarpic fruits as well as twisted leaf and stem axes. Additionally, the extent of elongation is positively linked to the level of SUN gene expression [4, 14]. Despite the fact that SUN has no substantial impact on fruit weight, it does influence tomato fruit morphology by enhancing longitudinal cell division and attenuating transverse fruit cell division [4]. Auxin performs an instrumental modulatory function in the regulation of cell division and expansion and cell identity establishment [17]. Microtubule dynamics were suggested to be influenced by auxin and to have roles in controlling post-embryonic division orientation [18] or cell shape [12]. Recent studies showed that IQD proteins may be involved in controlling cell shape or cell division by modulating auxin-mediated microtubule behavior [18]. Although plant phenotypes associated with elevated SUN expression levels indicated a role for auxin in controlling fruit shape [13], auxin level did not change dramatically in SUN as opposed to wild-type fruit. Further study found that SUN is linked to Ca2+ signaling and alters auxin signal transduction gene expression, demonstrating that SUN might influence fruit/ovary shape by modulating the auxin-associated gene expression level in the early phase of ovary formation [19]. Grape (Vitis L.) is amongst the most frequently produced fruit crops and it is characterized by a broad range of fruit sizes and shapes. Wild germplasm and wine grapes are usually circular or nearly circular. Modern domesticated table grapes have much more diverse shapes, including heart-shaped, ovoid, circular, narrow ellipsoid, nearly circular, obovoid, broad ellipsoid, and cylindrical [20]. The improvement of people’s living conditions has resulted in a greater emphasis on grape quality, both in the flavor and the aesthetic aspects. The cultivation and sale of fruit varieties with unusual fruit shapes have the potential to significantly increase economic advantages. Thus, the promotion of berry quality features, as well as the discovery of the genetic pathways that influence them, have gained considerable attention. To date, numerous QTLs and potential genes associated with berry weight and size have been genetically studied in grapevine [20]. However, research on berry shape mainly focuses on physiological aspects [1]; the genetic mechanism of this diversity remains unknown, although berries have been reported to exhibit a wide range of phenotypic diversity in shape. In this investigation, we discovered that VvSUN is a protein that has an IQ domain, localizes to the plasma membrane and chloroplast, and is highly expressed in the ovary 1 week before the anthesis stage. More importantly, the expression level is positively correlated with the fruit shape index (FSI). VvSUN overexpression in tomato and tobacco led to an elongated fruit/pod shape. We further showed that VvARF6 interacts with the promoter of VvSUN in yeast and that VvSUN activity was triggered by indole-3-acetic acid (IAA) treatment in vitro. Moreover, the IAA level and auxin-related genes were significantly altered in 35S::VvSUN transgenic tomato lines. Combined weighted gene co-expression network analysis (WGCNA) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis confirmed that the plant hormone signal transduction pathway may have an important role in controlling fruit shape. Our results offer a novel perspective on the function performed by VvSUN in modulating the elongated fruit shape in the auxin pathway during the early phases of fruit growth.
A BLASTP search of the Vitis vinifera genome using 33 Solanum lycopersicum SUN protein sequences as a query identified a total of 25 putative SUN genes (VvSUNs). Twenty-five genes were named VvSUN1–VvSUN25 according to their physical locations on the chromosomes. The phylogenetic tree was constructed with predicted S. lycopersicum SUNs, VvSUNs and other species’ SUN protein sequences using a neighbor-joining algorithm in MEGA11 with 1000 bootstrap replicates (Fig. 1A). VvSUN13 (LOC100253695), VvSUN14 (LOC100265924), and VvSUN18 (LOC100256816) were classified into the same subgroup as SlSUN1, indicating that these genes might share similar functions. SlSUN1 was identified as SUN, and its role in fruit shape control has been widely studied [14, 21]. Further study through PlantDGD (http://pdgd.njau.edu.cn:8080) online software revealed that VvSUN13 and VvSUN14 are a duplicate gene pair [22], both of them located on chromosome 8. Expression profiling of VvSUNs related to flowers and berries were analyzed by using the previously published grape (V. vinifera cv. ‘Corvinathe’) RNA-sequencing data from NCBI (accession number GSE36128) [23]. VvSUN13 showed high expression in young and well-developed inflorescences, the pericarp at mid-ripening stage, and berry skin at veraison stage, while VvSUN14 was expressed constantly from young inflorescences to flowering flowers and also in berries (skin, pericarp, and flesh) at post-fruit-set stage (Fig. 1B). These duplicate gene pairs that have diverged in their expression level indicated that their function might alter during evolution. VvSUN18 is not present in Fig. 1B because of the lack of corresponding expression data. VvSUN13, VvSUN14, and VvSUN18 were considered as the candidate genes in controlling fruit shape for further research.
Xiao et al. [14] discovered that the SUN transcription factor plays a role in the modulation of tomato fruit shape. Mature fruits harvested from six different grape cultivars were utilized to study the potential relationship between the level of VvSUN gene expression and fruit morphology (Fig. 2A), measure their longitudinal diameter and length (Supplementary Data Fig. S1A and B), and derive the FSI (length/diameter ratio) (Fig. 2B). The FSIs of ‘Gold Finger’ (GF),‘Minicure Finger’ (MF), and 8-6-1 (‘Beni Pizzutello’ seedling) were much higher than those of ‘Shine-Muscat’ (SM), ‘Kourgan Rose’ (KR), and ‘Houman’ (HM) (Fig. 2B). Different cultivars were used to assess the levels of VvSUN13, VvSUN14, and VvSUN18 mRNA transcripts in their ovary and young fruit growth phases at various periods (Fig. 2C, Supplementary Data Fig. S1C and D). VvSUN13 showed consistency in expression patterns among all the cultivars. It gradually increased in all cultivars 2 weeks before the anthesis (WBA) stage and reached the highest expression level at the 1 WBA stage; after that, the transcript levels were sharply decreased 3 days before the anthesis (DBA) stage and remained low until 2 weeks after the anthesis (WAA) stage. Moreover, the expression of VvSUN in GF, MF, and 8-6-1 at 1 WBA was much higher than in other cultivars. However, both VvSUN14 and VvSUN18 showed no direct relationship between the levels of VvSUN gene expression and fruit morphology (Supplementary Data Fig. S1C and D). Moreover, sequence comparisons revealed that VvSUN13 was most closely related to SlSUN, with 43.57% identity for the complete protein sequence. Therefore, we predicted the VvSUN13 gene as the corresponding grapevine SUN ortholog, which may also be involved in controlling fruit shape. Since VvSUN13 was first studied as VvSUN by Zhang et al. [24], we also refer to VvSUN13 as VvSUN in the present article. In accordance with these parameters measured above, the correlation coefficients between fruit shape parameters and the relative expression of VvSUN were calculated. The VvSUN expression level at the 1 WBA stage was positively correlated with the longitudinal length of the fruit and showed the highest correlation to FSI (0.87) (Fig. 2D). We additionally analyzed the expression profiles by fusing the VvSUN promoter region to the ß-glucuronidase (GUS) reporter (Supplementary Data Fig. S2). GUS staining was visible in the ovaries, anthers, stigmas, and young petals (Fig. 2E–I). These findings indicate that although VvSUN is predominantly expressed in the young ovaries, it may also function in other tissues, mostly those in which many cell divisions occur. A conserved domain search confirmed that VvSUN protein has the IQ67 domain, a conserved core section of 67 amino acids that are involved in the recruitment of calmodulin or function as a Ca2+ sensor. There are two distinct categories of IQ67 domains: one is the Ca2+-independent IQ motif, the IQ motif (I/L/VQxxxRxxxxR/K or IQxxxRGxxxR); the other one is the Ca2+-dependent IQ motifs, the 1-8-14 motifs (1-4 [FILVW]x6[FAILVW]x5[FILVW]) and 1-5-10 (1–4 [FILVW]x3[FILV]x4[FILVW]) [25]. The IQ67 domain of VvSUN contains two IQ motifs (amino acid residues 109–123 and 135–145), three 1-5-10 motifs, and two 1-8-14 motifs (Supplementary Data Fig. S3). Our results revealed that the elongated fruit shape was positively correlated with the high expression level of VvSUN at the 1 WBA stage, but what induced the VvSUN expression level variation is still unknown. To subsequently examine the genetic mechanisms for these differences in expression, we separately cloned and sequenced the cDNA and promoter sequences (from −1833 bp to ATG) of the VvSUN gene from these six grape cultivars. Sequence alignment analysis showed no direct link between the SNPs and fruit shape (Supplementary Data Figs S4 and S5), indicating that the VvSUN coding sequences and −1833 bp upstream cannot explain expression variation.
To verify the role of the VvSUN gene in the control of fruit shape, transgenic tomato lines were generated in which the VvSUN gene was overexpressed under the control of the cauliflower mosaic virus (CaMV) 35S promoter. From the five independent T4 generations of transgenic tomato lines that were produced, two VvSUN overexpression lines (lines 4 and 5) with the highest expression were selected for subsequent investigation (Supplementary Data Fig. S6A, Fig. 3A). The ovaries and fruits were obtained from the control and two overexpressing lines, and their longitudinal length, diameter, and shape index were determined (Fig. 3B and C, Supplementary Data Fig. S6B and C). The differences in length, width, and shape index were firstly observed in 1 WBA ovaries, and it was found that each genotype’s FSI did not alter much with the variation in days post-anthesis (DPA) and that it remained constant from 1 WBA till the mature stage (Fig. 3C). This shows that the shape of the fruit has already been decided during the early stages of ovary development. On the sliced paraffin section three distinct regions were identified: the pericarp, the columella, and the placenta (Supplementary Data Fig. S7). Cells in these regions steadily increased in size from 1 WBA to 5 DPA. Nonetheless, at any stage of development, the cell size and shape of transgenic tomatoes were comparable to those of the control (Fig. 3D). We further checked the cell numbers (number/mm2) in the pericarp both in longitudinal section and cross-section (Fig. 3E, Supplementary Data Fig. S8A), as well as cell shapes at 1 WBA stage, and the results showed no significant difference between transgenic tomatoes and control (Supplementary Data Fig. S8B–G), which implies that the elongated fruit morphology of lines 4 and 5 was mostly attributable to the creation of more cells in the longitudinal axis as a result of the increased rate of cell division. To further validate the function of VvSUN for elongated fruits, we also introduced 35S::VvSUN into tobacco (K326). Twelve putative transgenic lines were chosen on a medium that contained 30 mg l−1 hygromycin and confirmed by qRT–PCR (Supplementary Data Fig. S9A). Significant differences in pod morphology were discovered in tobacco plants that constitutively expressed VvSUN, with transgenic pods exhibiting a longer pod length, shorter pod width, and a higher pod shape index compared with pods from control plants (Fig. 3F and G, Supplementary Data Fig. S9). Interestingly, we also noticed that the transgenic tobacco plants had longer leaf rachises and an increased leaf shape index compared with the control plants. Cell forms and sizes in the lower epidermis leaves of transgenic tobacco were very similar to that in the wild type (Supplementary Data Fig. S5). Therefore, these findings also indicate that longer leaves contain more cells.
Numerous IQD genes in Arabidopsis are potential targets of ARF5, an early auxin-responsive factor, and the AtIQD15 expression level is elevated following exogenous auxin application [17]. Interestingly, in silico analysis of the cis elements present in the −1833-bp promoter sequence of the VvSUN genes revealed numerous motifs (Supplementary Data Table S1), including an ARFAT element that functions as an ARF binding site. Our previous research found that VvARF6 (LOC100242923) was activated by exogenous plant hormone treatment (not published data). To determine whether VvARF6 interacts with the VvSUN promoter, we examined the interactions between VvARF6 and the VvSUN promoter using a yeast one-hybrid (Y1H) assay. The Y1H results demonstrated that the VvARF6 protein interacted with the VvSUN promoter fragment, confirming that the VvARF6 protein recognizes the cis element in the VvSUN promoter in yeast (Fig. 4A). For the purpose of determining whether the VvSUN gene could be triggered by auxin, we transiently transformed the VvSUN promoter into grape leaves and measured the GUS activity in leaves treated with 0, 10, 50, and 100 mg/l of auxin. Compared with the mock control (35S::GUS), GUS activity mediated by the VvSUN promoter was significantly increased when exogenous auxin was applied, and the highest GUS activity was achieved at 50 mg/l IAA treatment (Fig. 4B and C).
We also transiently generated the GFP-VvSUN fusion protein controlled by the 35S promoter in Nicotiana benthamiana leaves and recorded its cell localization utilizing confocal laser scanning microscopy to establish the subcellular location where VvSUN operates. By overlapping the fluorescence of GFP and chlorophyll, strong fluorescence of the GFP-VvSUN fusion protein was identified in the plasma membrane and chloroplast (Fig. 4D).
Previous studies on IQD family proteins hypothesized possible links to auxin pathways [14], and the fruit shape phenotype of SUN-overexpressing plants is comparable to the shape of auxin mutants [4, 26]. In this study, auxin response factor VvARF6 interacted with the VvSUN promoter and induced GUS activity under different IAA treatments. To determine whether the VvSUN gene can regulate fruit shape by modulating auxin, UHPLC–MS/MS analysis was carried out to detect and quantify auxin and auxin-related compounds, such as IAA precursors (indole-3-acetamide, IPYA), free auxin (IAA), and IAA conjugates (indole-3-acetic acid-aspartate, IAA-Asp) in 35S::VvSUN line 5 and control at 1 WBA, anthesis and 5 DPA stages. VvSUN overexpression contributed to elevation in the levels of IPYA and IAA in all stages (Fig. 5A and B). IAA-Asp was also detected, but there was no significant difference between 35S::VvSUN line 5 and control at the 1 WBA and anthesis stages. However, the 35S::VvSUN line had significantly increased IAA-Asp at the 5 DPA stage, where the concentration was ~14.9-fold higher than in the control tomato (Fig. 5C). Collectively, the above findings imply that inactivation processes might be essential to maintain auxin homeostatic function and to prevent excessive IAA response amplification in cases where it has been initiated. In order to acquire a deeper comprehension of the auxin pathway, we investigated the transcriptional patterns of genes that could be involved in auxin biosynthesis, homeostasis, conjugation, and auxin influx transporter by RNA-seq at distinct stages of development in 35S::VvSUN and the wild type, as depicted in Fig. 5D and E. VvSUN significantly influenced the expression of 60 auxin-related genes, such as 3 auxin-biosynthesis-associated genes (TAA1, YUCCA5, and YUCCA10), 4 IAA-amino acid hydrolases (ILRs), 5 auxin homeostasis-related genes (GH3s), 8 polar transport genes (PINs and LAXs), and 40 signal transduction genes (ARFs, IAAs,SAURs, and TIR1-like gene) (Fig. 5D and E; gene_ID is listed in Supplementary Data Table S3). Regarding the interaction between auxin and VvSUN, we propose that VvSUN regulates tomato fruit shape not only by auxin levels but also by polar transport and/or auxin signal transduction processes.
To evaluate the mechanisms through which VvSUN modulates fruit shape, we determined the differentially expressed genes (DEGs) in pairwise comparisons of 35S::VvSUN transgenic tomato ovary/fruits and control tomato at different stages (1 WBA, anthesis, and 5 DPA stages). Analysis between VvSUN_1WBA versus control_1WBA, VvSUN_Anthesis versus control_Anthesis, and VvSUN_5DPA versus control_5DPA showed that 2971, 2118, and 2785 genes were upregulated, respectively, and that 3128, 1927, and 2919 genes were downregulated, respectively, at three different stages (Supplementary Data Figs S10 and S11). Gene ontology (GO) term enrichment (P ≤ .05) analysis illustrated that these DEGs were predominantly implicated in organic substance metabolism, primary metabolic process, oxidoreductase activity, catalytic activity, metabolic process, ribosome, and so on. In addition, genes related to transmembrane transport, calcium ion binding, cytoskeletal protein binding, tubulin binding, and microtubule-based movement were also enriched (Fig. 6A). Interestingly, we found that the expression of transmembrane transport pathway (GO:0055085)-related genes was significantly changed among the three stages (Supplementary Data Fig. S12; gene_ID is listed in Supplementary Data Table S4). We further clustered the DEGs on the basis of the log2-fold change in 35S::VvSUN and control utilizing the K-mean cluster. There were four patterns detected in the time-series gene expression profiles, which were then displayed utilizing a multigene line plot (Fig. 6B). The same DEG subclusters in two different genotypes showed different expression patterns and were considered the main genes regulated by the VvSUN gene. In subclusters 1 and 4, there were 317 and 217 DEGs, respectively, that exhibited contrasting expression trends. KEGG enrichment analysis showed a remarkable enrichment of these two subclusters in the plant hormone signal transduction pathway (Fig. 6C). In the GO analysis of clusters 1 and 4 (Supplementary Data Fig. S13), pathways were most related to the DNA metabolic process, cysteine-type peptidase activity, and mRNA maturation, such as spliceosomal snRNP assembly, SMN complex, ribonucleoprotein complex assembly, and ribonucleoprotein complex subunit organization.
In order to reveal gene networks associated with fruit shapes in 35S::VvSUN transgenic tomatoes, The gene expression profiles of all these 22 510 genes were analyzed to identify gene co-expression modules using the R package WGCNA. Here, 11 co-expression modules were identified, among which the ‘green’, ‘lightcyan’, ‘darkred’, and ‘pink’ modules were not only significantly associated with fruit shape but also with auxin-related compounds (Supplementary Data Fig. S14), indicating that auxin-related genes may be correlated with the control of fruit shape. Specifically, the longitudinal diameter was positively correlated with the expression of genes in the ‘green’, ‘darkred’, and ‘pink’ modules (Supplementary Data Fig. S14), with a coefficient of 0.77 (P = 2e−04), 0.83 (P = 2e−05), and 0.6 (P = 0.008), respectively (Supplementary Data Fig. S14). The transverse diameter was positively correlated with the expression of genes in the ‘brown’ and ‘green’ modules with a coefficient of 0.82 (P = 3e−05) and 0.57 (P = 0.01), respectively (Supplementary Data Fig. S14). Moreover, the FSI was positively correlated with the expression of genes in the ‘grey60’, ‘lightcyan’, ‘darkred’, and ‘pink’ modules with a coefficient of 0.5 (P = .04), 0.5 (P = .03), 0.63 (P = .005), and 0.79 (P = 9e−05), respectively. Genes clustered in above modules were picked out according to the gene significance (GS) values (GS > coefficient values of the trait) and P values (P.GS < P values of the trait) for further KEGG pathway analysis. The results showed that these genes were significantly enriched in several pathways (Supplementary Data Table S5). Interestingly, the plant hormone signal transduction pathway was a jointly owned pathway by longitudinal diameter, transverse diameter, and FSI trait modules (Fig. 7A). Then, genes involved in the processes of the plant hormone signal transduction pathway in each trait module were selected to construct a gene network by Cytoscape. As seen in Supplementary Data Fig. S15, out of the 41 hormone signal transduction pathway genes, 14 were plant hormone-related genes and among them around 50% were auxin-related genes according to the annotation of genes. Therefore, it is conceivable that the mechanisms underlying plant hormone signal transduction serve as the primary hub for interactions with other pathways implicated in controlling the elongated fruit morphology that arises from VvSUN overexpression in the plant.
The shape of the fruit is among the most distinguishing characteristics of the table grape. Nonetheless, only a few gene-oriented research reports have concentrated on the discovery of genes relevant for grape berry morphology, despite the fact that a vast spectrum of phenotypic diversity in berry shape has been reported. In this study, we cloned a grape VvSUN gene that encodes an IQD-like protein. The phylogenetic tree showed that VvSUN is one of the closest homologs to tomato SUN (Fig. 1A). Genes clustered in the same subclade are more likely to have similar functions [12]. Earlier research reports have demonstrated that upregulation of SUN contributed to the development of fruit that was very elongated and generally seedless [4]. Using anatomical findings, it was discovered that SUN has a significant influence on the morphology of the fruit before its anthesis, but it is after anthesis that SUN’s most striking influence on shape is evident, which is likely due to the changes in cell division rates in the longitudinal direction, leading to a cell number increase along the proximal–distal axis [4]. In our results, the expression levels of VvSUN were much higher in elongated grape cultivars than in round or near-round types at the 1 WBA stage (Fig. 2C). Moreover, the VvSUN expression level at the 1 WBA stage was positively correlated with the longitudinal length of the mature berry and also showed the highest correlation to FSI (0.87) (Fig. 2D). Overexpression of VvSUN driven by the 35S promoter in tomatoes led to an increase in the FSI in tomatoes but showed little or no impact on cell size and form (Fig. 3A–E, Supplementary Data Figs S6–S8). The function of VvSUN was also validated in transgenic tobacco, which showed a significant increase in the pod shape index as well as in the leaf shape index. Moreover, the cell form and sizes in lower epidermis leaves of transgenic tobacco were similar to those of the wild type (Fig. 3F and G, Supplementary Data Fig. S9). Given the difference in fruit types between tomato and tobacco, it is likely that the basic function of VvSUN in regulating the FSI by changing cell division is likely conserved. Furthermore, to our knowledge, VvSUN has not previously been recognized as a gene that regulates the morphology of fruits. Hence, we infer that VvSUN is a novel gene that modulates fruit shape by changing the cell division rate. Our data also show that the differences in the levels and timing of VvSUN expression were positively correlated with the fruit shape phenotype (Fig. 2C and D). However, sequence analyses revealed that there were no consensus sequence diversities in the coding regions and −1833 bp upstream of the VvSUN gene between the elongated and the round grape cultivars (Supplementary Data Figs. S4 and S5), indicating that the VvSUN coding sequences and −1833 bp upstream cannot be the reason for expression variation. Gene regulation in multicellular eukaryotes is complex, with many layers of regulation [27], including mutations in coding sequences or promoter regions [14, 28], long-range control by distant repressors or enhancers, alteration of epigenetic states, coordinated expression of genes [29], and regulation by transcription factors, including microRNAs (miRNAs), small interfering RNAs (siRNAs), messenger RNAs (mRNAs), and non-coding RNAs [30]. In this case, it is a big challenge and needs further effort in order to decipher how variation in regulatory mechanisms eventually results in changes in VvSUN gene expression profiles.
The plant hormone auxin performs a fundamental function in the modulation of cell expansion, cell division, and cell identity establishment [18, 31]. Earlier research reports illustrated that IQ domain-containing proteins belong to a calmodulin-binding protein family and play a role in the regulation of fruit shape by modulating auxin signal transduction [12, 32]. In this study, we also showed that the VvSUN promoter interacts with grape VvARF6 in yeast (Fig. 4A), and GUS activity driven by the VvSUN promoter was significantly increased when exogenous auxin was applied (Fig. 4B and C). More importantly, ectopic overexpression of VvSUN in tomatoes not only enhanced endogenous IAA content but also remarkably influenced the expression of auxin-associated genes, especially those implicated in polar transport and signal transduction (Fig. 5E). In addition, the combination of clustering, WGCNA, and KEGG enrichment analysis demonstrated a substantial enrichment of the DEGs in 35S::VvSUN transgenic tomatoes in the plant hormone signal transduction pathway in pairwise comparisons with control (Figs 6B and C and 7A). It has been shown that auxin signaling plays an important role in apple size [33] and the inhibition of polar auxin transport in tobacco (Nicotiana tabacum) changes the orientation of cell division [34, 35]. Recent studies have shown that multiple IQD genes are candidate ARF5 targets and are transcriptionally regulated by auxin signaling [12, 32]. Multiple studies have demonstrated that active auxin levels and distributions are closely regulated by the actions of synthesis, inactivation, and transport. Additionally, Wang et al. [13] detected that SUN shifted the expression of auxin polar transport and signal transduction during the initial stages of the ovary’s development. Thus, it is reasonably suggested that VvSUN alters the longitudinal direction of cell division in fruit by affecting auxin transport or the auxin signaling pathway. Recently, several lines of evidence have shown that auxin can influence microtubule dynamics [36], and genetically controlled microtubule depolymerization in embryos leads to the disruption of asymmetric divisions [18]. Most IQD members co-locate with microtubules, the cell nucleus, or membranes, which are implicated in the transduction of Ca2+ signals into cell responses via the modulation of a variety of target proteins [17, 37]. As a result of the elevation in cytosolic Ca2+ concentrations caused by auxin treatment, the activity of the IQD is regulated posttranslationally via stimulation of the Ca2+ CaM signaling pathway [38]. Ca2+ CaM, on the other hand, has an effect on auxin production by directly interfacing with components of the auxin transport and signaling mechanism, including PINOID (PID) or small auxin upmodulated RNA 19 (SAUR19) [38]. A conserved domain search confirmed that the VvSUN protein contains a Ca2+-dependent IQ motif (Supplementary Data Fig. S3), and we detected that VvSUN was localized in the plasma membrane and chloroplast (Fig. 4D). The transcriptome analysis of VvSUN overexpression in tomatoes revealed that DEGs were enriched in calcium ion binding, cytoskeletal protein binding, tubulin binding, and microtubule-based movement pathways (Fig. 6A). Moreover, the transmembrane transport pathway (GO:0055085) was activated in 35S::VvSUN transgenic tomato among the three stages using pairwise comparisons with control (Supplementary Data Fig. S12). The findings presented in this work lead us to postulate that VvSUN is situated in the plasma membrane and functions as a hub gene to translocate cellular auxin and calcium signaling (Fig. 7B). This, in turn, alters the pattern of cell division, contributing to the anisotropic expansion of the ovary and the presence of a fruit with an elongated morphology.
Grape plants (V. vinifera L.) of cv. ‘Minicure Finger’ (MF), ‘Kourgan Rose’ (KR), 8-6-1 (‘Beni Pizzutello’ seedling), and V. vinifera × Vitis labrusca cv. ‘GoldFinger’ (GF), ‘Houman’ (HM), and ‘Shine-Muscat’ (SM) from the Tang Shan Vineyard (College of Horticulture, Nanjing Agricultural University, Nanjing, China) were sampled during the 2020 growing season. Ovary and fruit samples were collected at 4 WBA, 3 WBA, 2 WBA, 1 WBA, 3 DBA, anthesis (when 50% of the caps were off), 3 days post anthesis (3 DPA), 1 WAA, and 2 WAA. The samples were promptly frozen in liquid nitrogen and kept at a temperature of 80°C till RNA extraction was done. Tomato (S. lycopersicum Mill. cv. ‘Micro Tom’) and tobacco (N. tabacum L. ‘K326’) plants were cultivated in a normal greenhouse environment at Nanjing Agricultural University. The following were the conditions under which the culture chamber was set up: 16-hour day/8-hour night cycle at constant 25°C, 60% humidity, and 250 μmmol m−2 s−1 luminous intensity [39].
The fruits of grapes (the six different cultivars mentioned above) and tomato, and the tobacco pod samples were measured for fruit/pod diameter and length by using a Vernier caliper (Mitutoyo, Kawasaki, Japan) at 7, 9, 15, 20, 36, 40 DPA. The ovaries and young fruits of tomato at 1 WBA, anthesis, and 5 DPA stages were measured with a stereo microscope (SZX10, Olympus, Japan). Histological examination of the fruits at various growth stages was accomplished by the use of paraffin segmentation. After fixing the obtained fruit samples in formaldehyde–acetic acid–ethanol (FAA) for at least 24 hours, they were washed in 50% ethanol for 10 minutes before being dried and embedded in paraffin in accordance with the conventional techniques reported by Godoy et al. [40]. The VvSUN expression level and phenotypic correlation coefficient were analyzed and visualized using online software (http://www.cloudtutu.com/).
The IQ67 domain of SlSUN [22] was utilized to categorize members of this family in grapes. Systematic BLAST screenings were conducted on all sequence data in genome databases for grape (version 2.0) and its annotation (http://ftp.ensemblgenomes.org/pub/plants/release-46/gtf/vitis_vinifera/) based on the domain retrieved from the SOL Genomics Network (SGN, http://solgenomics.net) as initial queries. The identified protein sequences were subsequently validated for IQ67 domain components in the SMART databases (http://smart.embl-heidelberg.de/) and NCBI Conserved Domains database (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). All the protein sequences containing errors or lacking the IQ67 domain were removed from the study. A phylogenetic tree was constructed utilizing the neighbor-joining (NJ) technique in the MEGA (Molecular Evolutionary Genetics Analysis) program version 11.0, and edge support was evaluated by applying 1000 bootstrap replicates [41]. The SUN sequences of additional species, such as cucumber (Cucumis sativus) and watermelon, were acquired from GuGenDB (http://cucurbitgenomics.org/) and the sequences of AtIQDs were retrieved from NCBI (https://www.ncbi.nlm.nih.gov/). This study comprised sequences with the following GenBank accession codes: V. vinifera, VvSUN1 (LOC100267340), VvSUN2 (LOC100260828), VvSUN3 (LOC100260747), VvSUN4 (LOC100254717), VvSUN5 (LOC100854442), VvSUN6 (LOC100254187), VvSUN7 (LOC100243111), VvSUN8 (LOC100240779), VvSUN9 (LOC100244856), VvSUN10 (LOC100266234), VvSUN11 (LOC100247418), VvSUN12 (LOC100256590), VvSUN13 (LOC100253695), VvSUN14 (LOC100265924), VvSUN15 (LOC100266890), VvSUN16 (LOC100241183), VvSUN17 (LOC100241471), VvSUN18 (LOC100256816), VvSUN19 (LOC100245132), VvSUN20 (LOC100245639), VvSUN21 (LOC104882151), VvSUN22 (LOC100255609), VvSUN23 (LOC100245291), VvSUN24 (LOC100263965), VvSUN25 (LOC100246947); Arabidopsis thaliana, AtIQD15 (AEE78534.1), AtIQD16 (AEE82912.1), AtIQD17 (AEE81939.1), AtIQD18 (AEE27239.1); C. sativus, CsSUN2 (XP_011659956.1); Citrullus lanatus, (Csa1G575000), C. lanatus, ClSUN8 (Cla011257); and S. lycopersicum, SlSUN1-SlSUN33 [21].
The Trizol® reagent (Thermo Fisher Scientific, USA) was employed to isolate total RNA in compliance with the guidelines stipulated by the manufacturer. From each sample, 1 μg of total RNA was obtained and subjected to treatment with RNase-free DNase (Vazyme, Nanjing, China) to eliminate any remaining genomic DNA, followed by conversion to cDNA with the aid of the First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, USA). Subsequently, we conducted qRT–PCR utilizing SYBR Premix ExTaq (Takara Biotech, Japan) on a Quant Studio™ 5 System (Thermo Fisher Scientific, USA). Normalization of the targeted gene was carried out by using the VvActin and SlActin genes. Next, the 2– ∆∆Ct method was employed to compute gene expression levels. As depicted in the corresponding figures, all qPCR tests for each biological replicate were carried out with three technical replicates. Supplementary Data Table S2 displays the primer sequences that were utilized for qRT–PCR.
Information on DNA and protein sequences was obtained from the NCBI database. The procedures reported by Zheng et al. [42] were utilized for DNA extraction, isolation of total RNA, first-strand cDNA synthesis, and DNase I treatment. The VvSUN (LOC100253695) cDNAs were isolated from six different grape cultivars during the pre-bloom phase by RT–PCR based on primers VvSUN-F/VvSUN-R (Supplementary Data Table S2) before cloning them into the pEASY®-Blunt Cloning Vector (TransGen). The VvSUN promoter (−1833 bp to ATG) from different cultivars was amplified by primers VvSUNpro-F/R from grape DNA, followed by cloning into the pEASY®-Blunt Cloning Vector (TransGen). Several clones were randomly chosen and verified by sequencing. The DNAMAN program (Lynnon Biosoft, San Ramon, CA, USA) was employed to assess sequence alignment. Supplementary Data Table S2 gives a complete list of all of the primer pairs that were utilized.
Amplification of the VvSUN gene’s coding sequence was accomplished utilizing PCR, followed by cloning of the resulting fragment into the pEASY®-Blunt Cloning Vector and subsequently into the pYH4215 vector utilizing a One Step Cloning Kit (Vazyme, Nanjing, China). A 35S::VvSUN construction was transformed into ‘Micro Tom’ tomato and tobacco via Agrobacterium-induced transformation (strain EHA105) in accordance with the procedures reported by De Jong et al. [43]. In half-strength MS media that contained hygromycin (30 mg l−1), potential transgenic lines were chosen, and their existence was subsequently verified by RT–qPCR, PCR, and GUS staining. Supplementary Data Table S2 shows the primers that were utilized in the PCR and RT–qPCR experiments. The T4 generation of the transgenic tomato and the T1 generation of the transgenic tobacco lines were chosen for physiologic experiments and molecular analyses.
The −1833 bp fragment (upstream from the start codon) obtained from the VvSUN promoter was amplified from grape genomic DNA and subsequently cloned into the pAbAi vector (Clontech) for the Y1H assay. Co-transformation of recombinant plasmid pGADT7-VvARF6 and pAbAi-VvSUN-pro into yeast strain Y1HGold (Clontech) was performed in accordance with the guidelines provided by the manufacturer. Additionally, transfection of the pGADT7 vector into baits was performed, which served as a negative control. SD/−Ura drop-out medium was used to culture the transformants. After a selection of colonies was made and dilution in sterile ddH2O attained an OD600 density of 0.5, 3 μl of suspension was spotted on SD/−Ura/−Leu drop-out containing AbA antibiotic at 30°C. Supplementary Data Table S2 provides detailed information on the primers that were utilized in this investigation.
The promoter of VvSUN was ligated into the pBI121-GUS vector before infusion into GV3101 to allow temporary expression in grape leaves. Subsequently, the grape leaves were grown in an incubator for 24 hours in darkness following vacuum infiltration and then treated with increasing dosages of 10, 50, and 100 mg/l of IAA 48 hours later. The positive and negative controls used in the experiment included the CaMV35S-GUS vector and the water treatment, respectively. GUS activity experiments were performed as previously described [44].
The VvSUN subcellular localization was determined by the PCR amplification of its full-length cDNA utilizing the primers VvSUN-GFP-F/R with incorporated NcoI and SpeI restriction regions and subsequent cloning into the pClone007 Blunt Simple vector (TSINGKE, China). Cloning of the full-length VvSUN cDNA into the pCAMBIA1302 vector was done after being verified by sequencing to generate the plasmid that would express the VvSUN-GFP fusion protein when driven by the 35S promoter. pCAMBIA1302-GFP was used as the control vector. The plasmids were added to Agrobacterium tumefaciens strain GV3101 following the protocol of Zheng et al. [42]. A Zeiss confocal scanning microscope (LSM700) was utilized to monitor the expression of the VvSUN-GFP fusion protein and chlorophyll signals in the infiltrated N. benthamiana leaves 3 days after inoculation.
IAA, IAA-Asp, and IPYA used as the standards were procured from Sigma–Aldrich (USA). Endogenous IAA and IAA-related compounds were extracted from ovaries and young fruits of tomatoes by using liquid–liquid extraction and determined by high-performance liquid chromatography-electrospray ionization tandem mass spectrometry (HPLC-ESI-MS/MS) following a published protocol [45]. Data were analyzed using MassHunter Workstation software (Agilent, CA, USA) and the final result was expressed in nanograms per gram FW.
We extracted total RNA from three biologically separate pools of the wild tomato (control) and 35S::VvSUN line 5 ovaries at 1 WBA, anthesis, and 5 DPA as described above in Expression investigations. The Stranded mRNA-seq kit (Vazyme, Nanjing, China) was utilized to create RNA-seq libraries. Next, the Illumina Novaseq platform (HiSeqTM2500/4000) was utilized for sequence analyses in Vazyme (China). A Trimmomatic (v0.33) was employed to screen the raw reads by eliminating the low-quality and adapter sequences. Mapping of the clean reads to the tomato genome was conducted by using STAR (v2.5.2b) [46]. DESeq (Padj <.05, v1.10.1) was employed to analyze the transcript assembly and expression levels of genes [47]. The DEGs affected by the genotype and developmental phase were clustered utilizing K-means in R [48]. Co-expression networks were created by using the WGCNA (v1.29) package in R and Cytoscape software (v3.9.1) [49]. The expression data used for WGCNA analysis comprised a total of 22 510 identified genes in the present study from the 18 datasets, and the trait data included longitudinal diameter, transverse diameter, FSI, IAA, IAA-ASP, IPYA, and fruit development stage. Clustering of the genes identified from K-means and WGCNA were performed using GO (GOSeq, v1.22) [50] and KEGG (KOBAS, v2.0) [51] enrichment analyses. The annotation file was downloaded from the tomato database (ftp://ftp.ensemblgenomes.org:21/pub/plants/release-47/fasta/solanum_lycopersicum/). Based on the gene’s annotations, transmembrane transport pathway genes and auxin-related genes implicated in its metabolic activities, signal transduction, and polar transport were identified from DEGs. Excel was utilized to perform pairwise comparisons of the expression values between the 35S::VvSUN and control at each developmental stage and the findings were subjected to log2 transformation.
This work was supported by the National Natural Science Foundation of China (Grant No. 31972384 and 31901975), China Agriculture Research System of MOF and MARA, the National Key Research and Development Program of China (2020YFD1000204), the China Postdoctoral Science Foundation (2019 M651858), the Jiangsu Agricultural Industry Technology System (JATS [2021]450), and the Jiangsu Key Agricultural Project for New Cultivars Innovation (PZCZ201723). The authors thank Proof-Reading-Service (https://www.proof-reading-service.com/) for language editing, which has improved the manuscript.
H.Z. and J.M.T designed the study. H.Z., Y.D., and H.L.N conducted the related experiments and data analysis and wrote the manuscript. J.L., X.Y., Y.G.Z., B.H., W.W., and L.Y.H. participated in the experiments. L.N.Y. and J.M.T. reviewed and revised the manuscript.
The RNA sequencing datasets generated in this study have been deposited in the National Genomics Data Center with the accession number PRJCA009128 (https://ngdc.cncb.ac.cn/). Other data supporting our findings are available in the manuscript file or from the corresponding author upon request.
The authors declare no competing interests.
Supplementary data is available at Horticulture Research online.
Click here for additional data file. |
||
PMC9647700 | Serena Matis,Anna Grazia Recchia,Monica Colombo,Martina Cardillo,Marina Fabbi,Katia Todoerti,Sabrina Bossio,Sonia Fabris,Valeria Cancila,Rosanna Massara,Daniele Reverberi,Laura Emionite,Michele Cilli,Giannamaria Cerruti,Sandra Salvi,Paola Bet,Simona Pigozzi,Roberto Fiocca,Adalberto Ibatici,Emanuele Angelucci,Massimo Gentile,Paola Monti,Paola Menichini,Gilberto Fronza,Federica Torricelli,Alessia Ciarrocchi,Antonino Neri,Franco Fais,Claudio Tripodo,Fortunato Morabito,Manlio Ferrarini,Giovanna Cutrona | MiR-146b-5p regulates IL-23 receptor complex expression in chronic lymphocytic leukemia cells | 14-07-2022 | Abstract Chronic lymphocytic leukemia (CLL) cells express the interleukin-23 receptor (IL-23R) chain, but the expression of the complementary IL-12Rβ1 chain requires cell stimulation via surface CD40 molecules (and not via the B-cell receptor [BCR]). This stimulation induces the expression of a heterodimeric functional IL-23R complex and the secretion of IL-23, initiating an autocrine loop that drives leukemic cell expansion. Based on the observation in 224 untreated Binet stage A patients that the cases with the lowest miR-146b-5p concentrations had the shortest time to first treatment (TTFT), we hypothesized that miR-146b-5p could negatively regulate IL-12Rβ1 side chain expression and clonal expansion. Indeed, miR-146b-5p significantly bound to the 3′-UTR region of the IL-12Rβ1 mRNA in an in vitro luciferase assay. Downregulation of miR-146b-5p with specific miRNA inhibitors in vitro led to the upregulation of the IL-12Rβ1 side chain and expression of a functional IL-23R complex similar to that observed after stimulation of the CLL cell through the surface CD40 molecules. Expression of miR-146b-5p with miRNA mimics in vitro inhibited the expression of the IL-23R complex after stimulation with CD40L. Administration of a miR-146b-5p mimic to NSG mice, successfully engrafted with CLL cells, caused tumor shrinkage, with a reduction of leukemic nodules and of IL-12Rβ1–positive CLL cells in the spleen. Our findings indicate that IL-12Rβ1 expression, a crucial checkpoint for the functioning of the IL-23 and IL-23R complex loop, is under the control of miR-146b-5p, which may represent a potential target for therapy since it contributes to the CLL pathogenesis. This trial is registered at www.clinicaltrials.gov as NCT00917540. | MiR-146b-5p regulates IL-23 receptor complex expression in chronic lymphocytic leukemia cells
Chronic lymphocytic leukemia (CLL) cells express the interleukin-23 receptor (IL-23R) chain, but the expression of the complementary IL-12Rβ1 chain requires cell stimulation via surface CD40 molecules (and not via the B-cell receptor [BCR]). This stimulation induces the expression of a heterodimeric functional IL-23R complex and the secretion of IL-23, initiating an autocrine loop that drives leukemic cell expansion. Based on the observation in 224 untreated Binet stage A patients that the cases with the lowest miR-146b-5p concentrations had the shortest time to first treatment (TTFT), we hypothesized that miR-146b-5p could negatively regulate IL-12Rβ1 side chain expression and clonal expansion. Indeed, miR-146b-5p significantly bound to the 3′-UTR region of the IL-12Rβ1 mRNA in an in vitro luciferase assay. Downregulation of miR-146b-5p with specific miRNA inhibitors in vitro led to the upregulation of the IL-12Rβ1 side chain and expression of a functional IL-23R complex similar to that observed after stimulation of the CLL cell through the surface CD40 molecules. Expression of miR-146b-5p with miRNA mimics in vitro inhibited the expression of the IL-23R complex after stimulation with CD40L. Administration of a miR-146b-5p mimic to NSG mice, successfully engrafted with CLL cells, caused tumor shrinkage, with a reduction of leukemic nodules and of IL-12Rβ1–positive CLL cells in the spleen. Our findings indicate that IL-12Rβ1 expression, a crucial checkpoint for the functioning of the IL-23 and IL-23R complex loop, is under the control of miR-146b-5p, which may represent a potential target for therapy since it contributes to the CLL pathogenesis. This trial is registered at www.clinicaltrials.gov as NCT00917540.
MicroRNAs (miRNAs) represent a family of noncoding RNAs that prevent the translation and promote the degradation of specific mRNAs by binding to their 3′-UTR., Several miRNAs have been implicated in the pathogenesis of chronic lymphocytic leukemia (CLL),3, 4, 5 a disease characterized by the accumulation of monoclonal CD5+CD19+ B cells in lymphoid organs and blood.6, 7, 8, 9 In patients with 13q deletions (del[13q]), the most common cytogenetic lesion of CLL,, the genes encoding the miR-15a/miR-16-1 cluster are targeted by the deletion.,12, 13, 14, 15 The downregulation of these regulatory miRNAs can lead to an increased expression of antiapoptotic molecules, which facilitate clonal expansion, inducing further transforming events.12, 13, 14, 15, 16 MiRNA expression profile studies have disclosed correlations between certain miRNA signatures and cytogenetic features and/or IGHV gene mutational status,17, 18, 19 which represent recognized prognostic markers of CLL. Finally, certain miRNA signatures are associated with disease progression and outcome,,20, 21, 22 or with the onset of a Richter transformation,23, 24, 25 a deadly condition characterized by the development of an aggressive lymphoma in CLL patients., Previously, we reported an inverse correlation between miR-146b-5p concentrations and progression-free survival in a cohort of >200 newly diagnosed Binet stage A patients; cases with the most aggressive clinical course had the lowest miR-146b-5p concentrations. The same inverse correlation was not observed with miR-146a-5p, a paralog of miR-146b-5p, in the same patient cohort. Although not validated by quantitative reverse transcription polymerase chain reaction (qRT-PCR), these differences were substantial and somewhat surprising, given that the 2 miRNAs share many predicted target genes and the same seed sequence. However, the 2 miRNAs are encoded by genes located on different chromosomes (chromosome 5 and 10 for miR-146a-5p and miR-146b-5p, respectively), which may create differences in the posttranscriptional processing associated with the 2 other nucleotides encoded at the 3′ end. Another surprising difference was that the CLL cases with the lowest miR-146b-5p concentrations were also IGHV-unmutated (UM), while this correlation was not observed in the case of miR-146a-5p. Both miR-146a-5p and miR-146b-5p control the proliferation of a variety of cells, particularly because they regulate NF-kB (nuclear factor kappa B) activation, a key transcription factor involved in cell proliferation., Both miR-146a-5p and miR-146b-5p exert a negative regulatory control on the expression of TNFR6 (tumor necrosis factor receptor-6) and IRAK1 (interleukin-1 receptor-associated kinase 1), 2 adaptor molecules that transduce signals delivered via several membrane receptors, such as those of the TNFR and the Toll-like receptor/IL1R superfamilies,30, 31, 32 culminating in NF-kB activation. This function accounts in part for the spontaneous onset of cancers in mice with deletions of miR-146a-5p, and the inverse correlation reported in human cancers between tumor aggressiveness and miR-146b-5p concentrations.35, 36, 37, 38, 39, 40 Inflammatory and autoimmune phenomena observed in mice with deletions of these miRNAs may also be explained by an absent NF-kB regulation.,41, 42, 43, 44, 45 However, the observation that miR-146b-5p is more effective than miR-146a-5p in determining CLL clinical course suggests that miR-146b-5p is implicated in additional mechanisms supporting CLL clonal expansion that are different from TRAF6 (tumor necrosis factor receptor-associated factor 6) and IRAK1 control. Considerable evidence indicates that CLL clonal expansion is promoted by interactions with cells and cytokines from the microenvironment., Moreover, both miR-146a-5p and miR-146b-5p can regulate the release of and the response to cytokines.,, Based on these considerations, we hypothesized that miR-146b-5p was involved in the regulation of the interactions between CLL cells and the microenvironment. We focused on IL-23, a cytokine of the IL-12 cytokine family, released primarily by dendritic cells, which is capable of driving T helper (Th) cell differentiation toward the Th17 cell subset. In a previous study, we found that IL-23 is instrumental in promoting CLL cell proliferation and clonal expansion. Normally, circulating CLL cells express variable concentrations of the IL-23R chain, 1 of the 2 chains forming the heterodimeric IL-23R complex, but are incapable of responding to IL-23 because of the absence of its complementary chain, IL-12Rβ1. Upon appropriate activation signals in vitro, such as the interaction with activated T cells or other CD40L-expressing cells, but not via direct stimulation of the B-cell receptor (BCR), CLL cells express the IL-12Rβ1 chain and begin to secrete IL-23. This initiates an autocrine/paracrine loop (which we have named the IL-23/IL-23R complex loop), whereby CLL cells respond to the IL-23 that they produce. This event promotes leukemic cell proliferation and appears to be very relevant for CLL cell growth/expansion since most leukemic cells in the proliferating centers of lymphoid tissues, infiltrated by CLL cells, produce IL-23 and express a complete IL-23R complex. Moreover, in vivo treatment with antibodies to IL-23p19 (1 of the 2 chains forming the IL-23 molecules) eradicates CLL clones in xenografted mice. Because the expression of the IL-12Rβ1 chain by CLL may represent a key checkpoint for the initiation of the loop, we hypothesized that miR-146b-5p was involved in regulating the expression of this chain. Indeed, the present findings support our hypothesis and show that miR-146b-5p can be a key regulator in controlling CLL cell clonal expansion.
The patients investigated were part of the O-CLL1 study (clinicaltrials.gov identifier NCT00917540), an observational cohort of patients with untreated Binet A CLL collected from several Italian institutions enrolled within 12 months from diagnosis., Supplemental Table 1 in the data supplement summarizes the clinical features of the patients investigated.52, 53, 54 In total, samples from 224 CLL cases were studied for expression profiles and single miR expression,; the data are deposited at the NCBI (National Center for Biotechnology Information) GEO (Gene Expression Omnibus) repository (http://www.ncbi.nlm.nih.gov/geo/) and are accessible through GEO Series accession number GSE40533. For CLL cases not included in the miRNome study, we measured miR-146b-5p concentrations by quantitative real-time PCR (RT-qPCR). Also, for these cases, miR-146b-5p expression was significantly correlated with immunoglobulin heavy chain variable region (IGVH) gene mutational status (see supplemental Methods, supplemental Table 2, and supplemental Figure 1). Peripheral blood mononuclear cells from patients with CLL were isolated by Ficoll-Hypaque (Seromed, Biochrom) density gradient centrifugation, and CD19-positive CLL cells were enriched by negative selection as previously reported (see supplemental Methods). Written informed consent was obtained from all patients in accordance with the declaration of Helsinki. The ethics committees from each participating center (listed in the acknowledgments) approved this study. Viable cell counts of CLL samples were conducted before each experiment performed in vitro and in vivo by trypan blue staining and automatic cell counter (Countess, Invitrogen). Values >80% of live cells were considered suitable for the subsequent experimental procedures.
MirVana miRNA mimics or inhibitors (Ambion Inc, Thermo Fisher Scientific; Grand Island, NY) were delivered to CLL cells using the Neon Transfection System (Invitrogen, Thermo Fisher Scientific) as described or by the Nucleofector-4D Transfection System (Amaxa), (supplemental Methods). The following miRNA mimics and inhibitors were used: hsa-miR-146b-5p (Assay ID: MC25960; MH25960), hsa-miR-146a-5p (Assay ID: MC10722), miRNA mimic, Negative Control#1 (Cat. no. 4464058), miRNA inhibitor, Negative Control #1 (Cat. no. 4464076). The transfection efficiency was verified by RT-qPCR (see supplemental Methods).
Cell surface IL-12Rβ1 and IL-23R chains were detected by flow cytometry. IL-12Rβ1 expression also was analyzed by Western blotting with mouse anti–IL-12Rβ1 monoclonal antibody (mAb) (C-20, sc-658, Santa Cruz Biotechnology, Inc.) and an anti-GAPDH mAb (AM4300, Ambion Inc, Thermo Fisher Scientific) as a loading control. qRT-PCR assessed the IL-12Rβ1 side chain mRNA (see supplemental Methods).
MiRNA target reporter vectors were purchased from Origene (IL-12Rβ1, Accession No. NM_153701, transcript variant 2, Cod. SC208722) and Switchgear (IL-23R, Cat. S806498, Accession No. NM_144701). IL-12Rβ1-MUT reporter vector, obtained by deletion of miR-146b-5p seed target site sequence (GTTCTCA [nt328-nt334]), was custom produced by Origene (Figure 2C). 3′ UTR assays are described in the supplemental Methods. HEK293 cells were used for transfection and the luciferase reporter assays. Preliminary tests showed that CLL-related cell lines (MEC-1 and OSU cell lines) were not suitable for testing because of the poor yield of the transfection step.
After transfection with the appropriate miRNA, CLL cells were cultured in RPMI 1640 medium with γ-irradiated cells from a stable CD40L-expressing NIH-3T3 (CD40L-TC) murine fibroblast cell line or with the NIH-3T3 cells transfected with the pIRES empty vector (Mock) (1 NIH-3T3 cell: 100 CLL cells) at a concentration of 2 × 106 cells per mL at 37°C in an atmosphere containing 5% CO2.
IL-23 cytokine production was measured in cell culture supernatants using the Human Cytokine/Chemokine Panel II and Luminex MAGPIX System (Merck Millipore).
These procedures were described previously,,, and additional details are provided in the supplemental Methods. All animal experiments were performed according to the current national and international regulations and were approved by the Licensing and Animal Welfare Body of the IRCCS-Ospedale Policlinico San Martino, Genoa, Italy.
The statistical package SPSS for Windows (release 13.0, 2004 software, SPSS UK; Surrey, United Kingdom) was used for all analyses. Statistical comparisons were performed using 2-way tables for the Fisher’s exact test and multiway tables for the Pearson’s χ2 test. Statistical comparisons between related samples were carried out by Wilcoxon or Mann-Whitney U tests. Time-to-first treatment (TTFT) analyses were performed using the Kaplan-Meier method. Statistical significance of associations between individual variables and survival was calculated using the log-rank test. The prognostic impact for the outcome variable was investigated by univariate and multiple Cox regression analysis. Data are expressed as hazard ratio (HR) and 95% confidence intervals (CIs). A value of P < .05 was considered significant for all statistical calculations. Values are given as mean ± SD.
First, we confirmed that miR-146b-5p concentrations maintained their prognostic power using a large CLL cohort described previously (O-CLL1 protocol). This comprised 224 Binet stage A patients, 48 of whom met the current diagnostic criteria of clinical monoclonal B-lymphocytosis., As shown in Figure 1A, miR-146b-5p was less expressed in CLL cases with IGHV-UM genes than in those with mutated IGHV (IGHV-M). The majority (41 of 56 [73%]) of cases with the lowest miR-146b-5p concentrations (first quartile) were IGHV-UM (Figure 1A). The median follow-up time in the cohort investigated was 83 months (range, 1-129), and 94 patients had progressed and required therapy at the time of the study censoring. Cases within the quartile with the lowest miR-146b-5p concentrations (first quartile) also had the shortest TTFT (Figure 1B). MiR-146a-5p failed to identify patients with a shorter TTFT (Figure 1C), a finding consistent with the observation that the concentrations of miR-146a-5p were similar in IGHV-M (n = 144, mean ± SD = 49 ± 95) and IGHV-UM cases (n = 80, mean ± SD = 45 ± 47) (supplemental Figure 2A-B). Furthermore, no correlation was observed between the expression of miR-146b-5p and miR-146a-5p, although the differences in expression between quartiles were similar for both miRNAs (supplemental Figure 2C-D). In a Cox multivariate model, together with other prognostic markers (IGHV-UM, CD38-positive, ZAP-70–positive, mutated NOTCH1 gene, RAI stage, FISH del(17p) or del(11q), β2-microglobulin (β2-M) values ≥5 mg/dL, and patients with a peripheral B-lymphocytosis of ≥5000/mm3), low miR-146b-5p expression failed to predict TTFT (supplemental Tables 3 and 4, Model 1). However, following the stratification of cases according to the IGHV mutational status, IGHV-UM cases with the lowest miR-146b-5p concentrations (first quartile) had TTFT curves that were significantly different from those of cases in the remaining quartiles. These differences were not observed in IGHV-M cases (Figure 1D-E). The analysis of the quartiles calculated within each IGHV-M and IGHV-UM group demonstrates the consistent survival association only within the IGHV-UM group (supplemental Figure 3). Cox multivariate analysis, with the variables used above, demonstrated a significant independent association between low concentrations of miR-146b-5p and clinical outcome (HR, 2.0; 95% CI, 1.1-3.9; P = .035) (Figure 1F) in IGHV-UM cases. Observations in 21 pairs of CLL cell samples taken from the same patients at disease onset and progression showed no changes in miR-146b-5p concentrations at disease progression (supplemental Figure 4).
To investigate possible mutations and copy number alterations (CNAs) on miR-146b and its putative promoter/enhancer regions possibly responsible for the occurrence of lower concentrations of miR-146b-5p in a subset of patients with CLL, we analyzed a dataset of 551 patients with CLL (CLLE-ES) by ICGC (International Cancer Genome Consortium) Data Portal (release_28), that collects sequencing data from different repositories, including the European genome–phenome archive. No patients with CLL presented somatic mutations in miR-146b genomic region (chr10:104196269-104196341) or putative promoter/enhancer regions predicted by GeneHancer (supplemental Table 5). CNA analysis performed on the same patients with CLL dataset showed the existence of a loss of the genomic region, including miR-146b, in 7 of 551 (1.3%) patients. CNA coordinates and patient characteristics are reported in supplemental Table 6. Similar results were obtained by Leeksma and colleagues, who retrospectively analyzed 2293 arrays for CNA assessment from 13 diagnostic laboratories according to established standards and found 10q losses in 25 of 2293 patients (approximately 1%). About half of these (13 of 2293 [0.6%]) showed 10q deletion encompassing miR-146b at the 10q24.32 locus. Therefore CNA at the miR-146b locus could not account for our observations. We then investigated the possibility that miR-146b-5p expression in CLL could be epigenetically regulated. Methylation status of miR-146b locus was explored in CLL cases, and normal B-cell samples by whole-genome bisulfite sequencing reported in the BluePrint Data Analysis portal (http://dcc.blueprint-epigenome.eu), considering a region spanning 500 bp upstream and downstream the miR-146b locus, respectively (GRCh37.p13 chr10:102436500-102436609, EnsEMBL version: 79). In CLL samples, mean concentrations of hypomethylation or emimethylation were detected in the region upstream or in the 150 bp immediately downstream miR-146b locus, whereas hypermethylation was found in the region encompassing miR-146b and in more downstream regions. In naïve and memory B cells from peripheral blood, respectively, a global pattern of hypomethylation was evidenced in the upstream regions, whereas hypermethylation was observed downstream to miR-146b locus (supplemental Figure 5). Therefore, a wider range and higher methylation concentrations than normal were observed in the upstream region of the miR-146b locus in CLL samples. In addition, the DNA methylation of the miR-146b-5p gene suggests that DNA methylation is directly involved in the regulation of its biogenesis. This regulation could be dependent on activating stimuli received by neoplastic cells in the lymphoid organs.
We next investigated whether miR-146b-5p could regulate the expression of IL-23R and/or IL-12Rβ1 chains. CLL clones can be subdivided into those with a low (IL-23R–low) or a high level of IL-23R (IL-23R–high) expression, respectively, when stratified according to a cutoff of IL-23R chain-positive cells lower or greater than 23%. We performed a correlation analysis to ascertain whether miR-146b-5p was lower in cases with higher IL-23R expression in a group of 93 CLL patients (40 cases IL-23R–low and 53 cases IL-23R–high). Although a significant anticorrelation in expression was detected (RHO−0.291; P = .005) (supplemental Figure 6), in vitro luciferase reporter assay failed to demonstrate a significant binding of miR-146b-5p to the IL-23R 3′UTR mRNA (Figure 2A). We used a recently developed web tool named miRabel (http://bioinfo.univ-rouen.fr/mirabel/) to investigate the potential binding of miR-146b-5p to the 3′UTR of the IL-12Rβ1 chain mRNA (for details, see supplemental Methods). This approach predicted a substantial binding capacity (score 0.3572489917218290), which was confirmed experimentally in luciferase reporter assays (Figure 2A), showing an average reduction of the luciferase activity of 35 ± 10.3% (mean ± SD). In contrast, miR-146b-5p did not efficiently bind the IL-12Rβ1 3′UTR–MUT with an average reduction of the luciferase activity of 10 ± 6% vs 33 ± 11% of the 3′UTR WT (mean ± SD; P = .01) (Figure 2C-D) confirming the specificity of the interaction between the miR-146b-5p seed sequence and the complementary sequence on the IL-12Rβ1 chain mRNA. The possible binding of miR-146a-5p to the 3′UTR of the IL-12Rβ1 chain mRNA, predicted by the same algorithms, was not confirmed experimentally (Figure 2). To further confirm the miRNA-mediated regulation IL-12Rβ1 side chain, primary CLL cells were transiently transfected with a specific miRNA inhibitor targeting miR-146b-5p or with a miR-control inhibitor (a random sequence molecule with no identifiable effects on known miRNA functions) and cultured for 48 hours. A consistent upregulation of the IL-12Rβ1 side-chain protein by knocking down miR-146b-5p expression was found by Western blot (Figure 2E-G).
The above target validation experiments prompted tests aimed at verifying whether miR-146b-5p inhibition could induce the expression of the Il-12Rβ1 side chain and a functional IL-23R complex on the surface of CLL clones already expressing an IL-23R side chain. Purified CLL cells from 8 IL-23R–high cases (35 ± 11% [mean ± SD] positive cells) (GE1-AG114, GE1-DM210, GC0015, SV1-SA, SR1-ME1077, MG0482, VF0384, and CM18) were transiently transfected with a miR-146b-5p inhibitor or with a miR-CTR inhibitor, cultured for different times, and tested for IL-12Rβ1 and IL-23R side chains expression. Cells transfected with miR-146b-5p inhibitor had a significantly increased expression (P = .0078) of IL-23R complex (average increase value of 57 ± 12% positive cells at 72 hours in culture [mean ± SD]) compared with the control cells (Figure 3A-B). The increased IL-23R complex expression was associated with an upregulation of the IL-12Rβ1 side chain (average increase of 51 ± 9% positive cells at 72 hours [mean ± SD]) that was significantly different (P = .0078) from the control samples; in contrast, the expression of the IL-23R side chain remained virtually unchanged (Figure 3C-D). The cells positive for the chains of the IL-23R complex were identified within the gated populations of viable cells (Figure 3A and supplemental Figure 7). IL-23 released by CLL cells in the culture supernatants was also measured. There were no differences in the IL-23 produced by the miR-146b-5p inhibitor transfected and control cells (Figure 3E). Next, we investigated the functional features of the IL-23R complex expressed by CLL cells. CLL cells purified from the same 8 patients were transiently transfected with the miR-146b-5p inhibitor or with the miR-CTR inhibitor and cultured in the presence or absence of recombinant IL-23 for different time points. Upon exposure to exogenous IL-23 in culture, significant increases (P = .0078) in cell viability (mean ± SD increase at 72 hours, 22.6 ± 10%) (Figure 3F-G) and of cycling cells (mean ± SD increase, 48.7 ± 27%) (Figure 3H-I) were observed in suspensions treated with the miR-146b-5p inhibitor; these effects were abrogated by the addition of a specific IL-23 mAb (αIL-23p19) to the culture supernatant (average inhibition at 72 hours, 37 ± 16%, for cell viability, and 84 ± 15% for cycling cells) (Figure 3F-I).
Since stimulation of CLL cells with CD40L in vitro induces the expression of a functional IL-23R complex, we investigated whether the same stimulation caused the downregulation of miR-146b-5p. Purified CLL cells from the 8 patients studied above were either transfected with the miR-146b-5p inhibitor or cultured with CD40L-TC. In both instances, the expression of the IL-12Rβ1 chain (and consequently of the IL-23R complex) was observed in amounts superior to those observed in the respective control cultures (average increase of 46 ± 19% for miR-146b-5p inhibitor treatment and of 77.3 ± 15.6% for CD40L-TC stimulation [mean ± SD], respectively) (Figure 4A-B). To investigate the concentrations of miR-146b-5p following CD40L-TC stimulation, purified CLL cells from 3 different cases with a variable baseline amount of miR-146b-5p were cultured with CD40L-TC and harvested at intervals. Viable cells were measured by flow cytometry by excluding annexin-V/PI-positive cells, whereas activated cells were identified as CD80+ cells (Figure 4C). Cell viability remained high throughout the culture, while there was a progressive acquisition of CD80 expression over time. Exposure to CD40L in vitro caused a substantial downregulation of miR-146b-5p as assessed by RT-qPCR (average inhibition at 48 hours, 70.1 ± 1.7% [mean±SD]) (Figure 4C). In contrast, stimulation of purified CLL cells by coculture with anti-μ and anti-δ Ig-chain–coated beads and IL-4 failed to significantly modify miR-146b-5p expression (supplemental Figure 8 and supplemental Methods). These data also are consistent with our previous findings on the incapacity of cell stimulation via BCR to induce the IL-23R complex expression. Next, purified CLL cells were transfected with the miR-146b-5p inhibitor or to the miR-CTR inhibitor for 6 hours, stimulated with CD40L-TC for 48 hours, and the concentrations of IL-12Rβ1 mRNA determined by RT-qPCR. Preexposure to the miR-146b-5p inhibitor caused a substantial increase of intracellular IL-12Rβ1 mRNA (average increase of 46.5 ± 40% [mean ± SD]) compared with the control samples (Figure 4D) irrespective of the baseline values of miR-146b-5p expression and with a wide variability depending on the propensity to CD40L activation of the different CLL clones. Flow cytometry tests confirmed these observations. Following pretreatment with the miR-146b-5p inhibitor, there was a consistent increase of IL-12Rβ1 (average increase of 25.6 ± 14.6% positive cells [mean ± SD]) and IL-23R complex expression (average increase of 36.4 ± 11.5% [mean ± SD]) compared with control samples (Figure 4E-F). Notably, pretreatment of the purified CLL cells with miR-146b-5p inhibitor before coculturing with CD40L-TC did not cause upregulation of the IL-21R, indicating a selective regulation of the miR-146b-5p on IL-23R complex expression (Figure 4G-H).
If miR-146b-5p concentrations regulate the IL-12Rβ1 expression, then a forced increase of intracellular miR-146b-5p should prevent the expression of IL-12Rβ1 following coculture with CD40L-TC. To test this, purified CLL cells from 10 different cases with different baseline miR-146b-5p expression (Figure 5 and supplemental Table 2) were cultured with miR-146b-5p mimic or miR-CTR mimic for 6 hours, CD40L-TC was added, and the cultures continued for 48 hours. Following transfection with miR-146b-5p mimics, lower IL-12Rβ1 mRNA concentrations were detected by RT-qPCR (Figure 5A). Flow cytometry confirmed that transfection with miR-146b-5p mimic prevented the expression of the surface IL-23R complex expression mediated by CD40L activation (average decrease of 55.4 ± 18% [mean ± SD]) (Figure 5B-C) mainly caused by surface downregulation of IL-12Rβ1 side chain (average decrease of 45.2 ± 22 [mean ± SD]) (Figure 5D). Concomitantly, there was a slight decrease in the overall expression of the IL-23R chain (average decrease of 19 ± 16 [mean ± SD] (Figure 5E), while the CLL cells expressing the IL-23R side chain only were increased (average increase of 39 ± 28 [mean ± SD]) (Figure 5F). Finally, a 50 ± 11.5% (mean ± SD) decrease of Ki67+ cells (Figure 5G,I) and of Ki67+ cells expressing the IL-23R complex (Figure 5H) compared with control samples was observed. Pretreatment of the purified CLL cells with miR-146b-5p mimic before coculture with CD40L-TC failed to cause upregulation of the IL-21R, as shown in supplemental Figure 9, confirming a selective effect of the miR-146b-5p mimic. Since miR146b-5p is known to repress TRAF6 and IRAK-1, which play critical roles in NF-kB activation,,, we investigated whether enforced expression of this miRNA caused downregulation of these targets in CLL cells. Since miR146a-5p has similar effects, the 2 miRNAs were tested in parallel. CLL cells were exposed to miR-146a-5p or miR-146b-5p or miR-CTR mimics for 6 hours in vitro and subsequently cocultured with CD40L-TC or mock cells for 48 hours. As shown in supplemental Figure 10A-B, TRAF6 protein was downregulated in CLL cells transfected with either miR-146a-5p or miR-146b-5p mimics compared with the control samples. Some TRAF6 inhibition, although at lower concentrations, was observed in control cocultures with mock cells. Similar results were obtained when IRAK1 expression was tested by flow cytometry in the same culture settings (supplemental Figure 10C-D).
Cells from GE1-PM129 and GE1-RO148 CLL cases were cocultured with activated autologous T cells and used to generate xenografts in 10 and 4 NSG mice, respectively. After 4 to 6 weeks, all mice presented circulating human (CD45+CD19+CD5+) cells indicative of successful engraftment. The mice were subdivided into equal groups, and each group of animals was treated with either miR-146b-5p mimic or miR-CTR mimic (1 injection on alternate days for a total of 3 injections). Flow cytometry analyses of samples from peripheral blood, bone marrow, and spleen cells, 3 days after the last miRNA injection, revealed that mice treated with miR-CTR mimic had higher percentages of CD45+CD19+CD5+ CLL cells and lower percentages of CD45+CD19−CD5+ T cells than mice treated with miR-146b-5p mimic (Figure 6A-C and supplemental Table 7). Mice treated with miR-146b-5p mimic presented a higher percentage of apoptotic (annexin-V–positive) CLL cells in the spleen (Figure 6D-E). This finding was consistent with the in situ immunohistochemical (IHC) analysis showing a decrease of spleen infiltration by leukemic (human CD20+) cells after treatment with miR-146b-5p mimics (Figure 6F). Autologous T cells (human CD3+ cells), surrounding remnants of CLL infiltration foci, were still present (Figure 6E-F and supplemental Table 7). In addition, the boundaries of the follicles appeared less evident and were often disrupted by the accumulation of T cells (Figure 7A-B). Engraftment was measured by determining an IHC index derived from the combination of size and numbers of CD20+ follicles in the spleen (supplemental Methods). A significantly lower IHC index was observed in mice treated with the miR-146b-5p mimic compared with control mice (127.6 ± 51 vs 34 ± 31.4 [mean ± SD]; P = .007) (supplemental Table 7). Moreover, in mice treated with the miR-146b-5p mimic, there were fewer Ki67+ cells in the spleen infiltrates (Figure 7C). Double-marker IHC confirmed the presence of fewer cycling CLL cells (human Ki67/CD20+) in the spleen infiltrates of mice treated with the miR-146b-5p mimic compared with the control samples (Figure 7D). Furthermore, staining with specific antibodies showed fewer IL-12Rβ1–expressing cells in mice treated with miR-146b-5p mimic than in the control samples (Figure 7E-F). Notably, a consistent number of cells present in the CLL cell aggregates were stained by anti–IL-23 mAb, indicating that the miR146b-5p mimic treatment did not affect IL-23 cytokine production (Figure 7F). Likewise, there were numerous T cells in the tissues analyzed, indicating that the T-cell compartment was not prominently affected by miR-146b-5p mimic treatment (Figure 7E).
The idea for this study stemmed from the consideration that miR-146b-5p had a relevant prognostic impact in CLL and that the IL-23/IL-23R complex loop is important for promoting CLL cell clonal expansion. Since IL-12Rβ1 is expressed following cell activation, this step may represent a relevant checkpoint for the functioning of the loop, and miR-146b-5p could conceivably determine the cell’s susceptibility to IL-23 by regulating IL-12Rβ1 receptor expression. The collected evidence supports the working hypothesis: miR-146b-5p proved capable of binding to the IL-12Rβ1 chain mRNA in an in vitro luciferase assay, whereas miR-146a-5p failed despite sharing the same seed sequence. A partial explanation for this failure could be that the binding of miRNAs associated with the argonaute protein to the relevant mRNA is influenced by sequences flanking the binding sites and by additional noncanonical binding sites. Thus, small sequence variations outside the seed sequence, and the different posttranslational processing of the 2 miRNAs, may cause variations in their binding to target mRNA. The capacity of miR-146b-5p to regulate IL-12Rβ1 expression was confirmed by experiments with specific miR-146b-5p mimics and inhibitors because the former prevented and the latter promoted IL-12Rβ1 expression. This effect was selective given that the expression of IL-21R, which also plays an important role in regulating CLL cell expansion,, was unaffected by miR-146b-5p. Notably, miR-146b-5p did not bind to IL-23 mRNA and did not appear to influence IL-23 production by CLL cells, indicating that IL-12Rβ1 chain expression is a major regulatory step in the IL-23/IL-23R complex loop. NSG mice engrafted with CLL cells and treated with miR-146b-5p mimic presented a reduction of both circulating and tissue-infiltrating leukemic cells compared with mice treated with CTR mimics and a disruption of the leukemic nodules, which had less defined boundaries, inferior numbers of proliferating cells, and appeared infiltrated by T cells. The expression of the IL-12Rβ1 chain by leukemic cells was markedly diminished. The concentrations of human T cells remained apparently unaltered in the engrafted NSG mice upon miR-146b-5p mimic administration, indicating that the treatment did not influence T-cell viability in this setting, although previous reports described a regulatory function of miR-146b-5p in follicular Th cells and regulatory T cells., Whether the T-cell subset distribution is altered remains to be ascertained. The activation of the IL-23/IL-23R complex loop in CLL cells is achieved mainly by stimulation through the surface CD40-dependent, not the BCR-dependent pathway. CD40 is a member of the TNFR family and requires interaction with TRAF6 and IRAK1 to activate NF-kB., Elevated concentrations of miR-146a-5p and of miR-146b-5p cause downregulation of IRAK1 and TRAF6 (supplemental Figure 10) in CLL, in principle, rendering NF-kB activation and stimulation via surface CD40 less effective.,, However, the observation that miR-146a-5p, which downregulates TRAF6 and IRAK1 expression as efficiently as miR-146b-5p, was not associated with prognosis in CLL suggested that the IL-23/IL-23R complex loop had a more critical role in regulating CLL cell growth. Interestingly, mice in which miR-146a-5p or miR-146b-5p is knocked out (KO) both develop lymphomas, although only the lymphomas originated in miR-146b-5p KO mice present a resemblance to human CLL. The reasons for this hierarchy in the mechanisms regulating CLL clonal expansion are far from clear. One possibility is that additional signals delivered by surface molecules different from CD40 and not requiring TRAF6 and IRAK1 adaptors are involved in activating the IL-23/IL-23R loop in vivo. An alternative and not mutually exclusive option could be offered by the redundancy of the TRAF/IRAK family members, whereby other molecules of the same families could substitute for the downregulation of TRAF6 and IRAK1 induced by the miR-146a/b., Notably, other miRNAs can regulate IL-23 stimulatory signals. This is the case of miR-221 and miR-222 that negatively regulate the susceptibility of Th17 cells to IL-23 stimulation by modulation of the IL-23R complex. The issue as to why CLL clones are heterogeneous in the miR-146b-5p concentrations is presently unclear, although it could be related to the different states of activation of the cells from the different CLL clones. This hypothesis is supported by the observations that CLL cell activation with CD40L-TC causes downregulation of miR-146b-5p concentrations in vitro and that the lowest miR-146b-5p concentrations are detected in IGHV-UM cases whose leukemic cells are at the highest activation status determined by surface marker analysis., An alternative and not mutually exclusive hypothesis poses that lesions of the miR-146b gene or regulatory DNA sequences facilitate the maintenance of low miR-146b-5p concentrations in the most aggressive CLL clones. However, this hypothesis is made unlikely by the finding that CNAs were very low in the database analysis we have carried out, and virtually no mutations of the miR-146b-5p locus are detectable in the same database., Alterations in the methylation of the miR-146b-5p locus of CLL cells compared with normal cells have been noticed and are reported in the Blueprint data analysis portal (supplemental Figure 5), a finding that could at least in part explain the heterogeneity of miR-146b expression in CLL, possibly dependent on the activation status of the neoplastic clones. This issue is currently being investigated. The present study has translational relevance as it indicates miR-146b-5p is a potential target through which the susceptibility of CLL cells to IL-23 could be modified. Future studies should investigate a strategy based on increasing intracellular miR-146b-5p concentrations as an application for CLL therapy, either alone or combined with anti–IL-23 mAbs,73, 74, 75 in the attempt to eradicate CLL, which so far has proven virtually incurable.
Conflict-of-interest disclosures: The authors declare no competing financial interests. |
Subsets and Splits